One scale, fit over other people's scores.
The openly licensed boards on /benchmarks measure different things on scales that do not convert, so none of them is added to another. This is the other way to put them on one axis: a capability per run and a difficulty per benchmark column, fit together, so both sit on the same rule. Vitna ran none of the evaluations behind it.
319 ranked of 521 runs · 46 columns · 3,630 cells · 15.2% full·boards read ·fit
12 of 46 columns sort runs within one standard deviation of their difficulty; 5 are wider than the whole ranked range and sort nothing on it.
- 01FrontierMath-ErdosEpoch AI Benchmarking Hub · 153 on the scale · above every ranked run
- 02SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub · 139 on the scale · above every ranked run
- 03SpireClimbEpoch AI Benchmarking Hub · 137 on the scale · above every ranked run
- 04SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub · 136 on the scale · above every ranked run
- 05spatialLiveBench · 53 on the scale · below every ranked run
- 06AMPS_HardLiveBench · 47 on the scale · below every ranked run
- 07math_compLiveBench · 42 on the scale · below every ranked run
- 08tablereformatLiveBench · 41 on the scale · below every ranked run
The ranked runs span 61 to 136. 8 columns sit outside that: 4 sit above every ranked run, where nobody yet has an even chance, and 4 sit below every ranked run, where almost everything passes. The dashed outline is the 202 runs set aside for standing on fewer than 3 columns, kept in the fit and out of the ordering; 0 of them sit beyond the right edge. Runs are counted in bins of 5 points. A column in the accent sorts runs within one standard deviation of its difficulty; the rest are the field. Each numbered column is a door to its row below.
001gpt-6.2-astranot in the catalogmax135.84141
14 columns on 1 board, hardest first are the whole evidence behind 135.84. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.10153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.33139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.41137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.49136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.92123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.95121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.89120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.78119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.68119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub1116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.79115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.93109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9783 on the scale
002gemini-4-neonnot in the catalogxhigh134.5481
8 columns on 1 board, hardest first are the whole evidence behind 134.54. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.85121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.89120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.83119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.98116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.75115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.91109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9683 on the scale
003openai/gpt-6-astramax132.41372
37 columns on 2 boards, hardest first are the whole evidence behind 132.41. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.03153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.33139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.34137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.35136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.76123 on the scale
- typescriptLiveBench43.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.84121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.80120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.74119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.60119 on the scale
- tablejoinLiveBench58.37117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.98116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.76115 on the scale
- pythonLiveBench65109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.94109 on the scale
- plot_unscramblingLiveBench86.29106 on the scale
- integrals_with_gameLiveBench100104 on the scale
- consecutive_eventsLiveBench90.56102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8896 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench70.5392 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench78.2586 on the scale
- summarizeLiveBench79.5585 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9683 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench73.9874 on the scale
- olympiadLiveBench92.1772 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
004claude-fable-5-2not in the catalogmax132.06121
12 columns on 1 board, hardest first are the whole evidence behind 132.06. Each is a door to the column and everything else scored on it.
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.13139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.15137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.17136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.90123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.90121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.90120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.79119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub1116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.94115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.95109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub1.0094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
005google/gemini-4-carbonhigh131.68101
10 columns on 1 board, hardest first are the whole evidence behind 131.68. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.03153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.88123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.73121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.78120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.78119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.95116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.78115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.91109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9783 on the scale
006openai/gpt-6.1-terramax130.85121
12 columns on 1 board, hardest first are the whole evidence behind 130.85. Each is a door to the column and everything else scored on it.
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.10139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.22137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.35136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.78123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.74121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.79120 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.77119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.93116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.57115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.91109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
007openai/gpt-6.1-solmax130.00372
37 columns on 2 boards, hardest first are the whole evidence behind 130.00. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.04153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.18139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.24137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.29136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.54123 on the scale
- typescriptLiveBench40122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.80121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.80120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.68119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.61119 on the scale
- tablejoinLiveBench57.42117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub1116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.74115 on the scale
- pythonLiveBench60109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.94109 on the scale
- plot_unscramblingLiveBench84.38106 on the scale
- integrals_with_gameLiveBench100104 on the scale
- consecutive_eventsLiveBench90.56102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench71.9792 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench72.5886 on the scale
- summarizeLiveBench81.2585 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9583 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench70.8274 on the scale
- olympiadLiveBench92.2772 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
008claude-opus-5-5not in the catalogmax128.09372
37 columns on 2 boards, hardest first are the whole evidence behind 128.09. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.03153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.05139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.06137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.07136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.71123 on the scale
- typescriptLiveBench63.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.71121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.83120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.67119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.96119 on the scale
- tablejoinLiveBench52.33117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.95116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.72115 on the scale
- pythonLiveBench70109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.91109 on the scale
- plot_unscramblingLiveBench77.48106 on the scale
- integrals_with_gameLiveBench100104 on the scale
- consecutive_eventsLiveBench90.56102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench81.8299 on the scale
- logic_with_navigationLiveBench8696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench58.6792 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench86.9687 on the scale
- paraphraseLiveBench64.2786 on the scale
- summarizeLiveBench69.1085 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench91.5576 on the scale
- story_generationLiveBench70.9274 on the scale
- olympiadLiveBench93.2572 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
009google/gemini-4-hydrogenxhigh125.83121
12 columns on 1 board, hardest first are the whole evidence behind 125.83. Each is a door to the column and everything else scored on it.
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.10139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.08137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.05136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.70123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.65121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.57120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.60119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.88116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.71115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.88109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9783 on the scale
010claude-opus-5-5not in the catalogxhigh125.14231
23 columns on 1 board, hardest first are the whole evidence behind 125.14. Each is a door to the column and everything else scored on it.
- typescriptLiveBench53.33122 on the scale
- tablejoinLiveBench51117 on the scale
- pythonLiveBench70109 on the scale
- plot_unscramblingLiveBench77.16106 on the scale
- integrals_with_gameLiveBench99104 on the scale
- consecutive_eventsLiveBench90.37102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench8096 on the scale
- simplifyLiveBench6392 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench86.9687 on the scale
- paraphraseLiveBench64.6886 on the scale
- summarizeLiveBench69.3385 on the scale
- typosLiveBench8077 on the scale
- code_generationLiveBench91.5576 on the scale
- story_generationLiveBench71.1874 on the scale
- olympiadLiveBench92.1572 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
011claude-fable-5-1not in the catalogmax125.12362
36 columns on 2 boards, hardest first are the whole evidence behind 125.12. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.02137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.04136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.57123 on the scale
- typescriptLiveBench60122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.58121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.70120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.48119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.93119 on the scale
- tablejoinLiveBench56.15117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.88116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.71115 on the scale
- pythonLiveBench70109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.90109 on the scale
- plot_unscramblingLiveBench75.17106 on the scale
- integrals_with_gameLiveBench99104 on the scale
- consecutive_eventsLiveBench90.56102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench8696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench70.1792 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench71.5286 on the scale
- summarizeLiveBench72.1785 on the scale
- typosLiveBench9477 on the scale
- code_generationLiveBench90.1476 on the scale
- story_generationLiveBench78.1074 on the scale
- olympiadLiveBench92.9772 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench94.1241 on the scale
012google/gemini-4-argonhigh124.90111
11 columns on 1 board, hardest first are the whole evidence behind 124.90. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.01153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.65123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.64121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.72120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.59119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.59119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.78116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.76115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.86109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9683 on the scale
013openai/gpt-6.1-solxhigh124.64231
23 columns on 1 board, hardest first are the whole evidence behind 124.64. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- tablejoinLiveBench55.25117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench81.94106 on the scale
- integrals_with_gameLiveBench99104 on the scale
- consecutive_eventsLiveBench91.29102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8296 on the scale
- simplifyLiveBench68.1792 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench71.0386 on the scale
- summarizeLiveBench75.5385 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench70.5774 on the scale
- olympiadLiveBench91.8672 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
014openai/gpt-6-solmax124.00372
37 columns on 2 boards, hardest first are the whole evidence behind 124.00. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.01153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0.05139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.09137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.13136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.53123 on the scale
- typescriptLiveBench40122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.56121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.58120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.56119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.49119 on the scale
- tablejoinLiveBench54.02117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.90116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.61115 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.90109 on the scale
- plot_unscramblingLiveBench77.90106 on the scale
- integrals_with_gameLiveBench99104 on the scale
- consecutive_eventsLiveBench89.55102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench69.2892 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench70.0886 on the scale
- summarizeLiveBench6885 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench83.1076 on the scale
- story_generationLiveBench66.9374 on the scale
- olympiadLiveBench91.3672 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
015openai/gpt-5.6-solmax123.35332
33 columns on 2 boards, hardest first are the whole evidence behind 123.35. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.45123 on the scale
- typescriptLiveBench50122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.58121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.57120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.59119 on the scale
- tablejoinLiveBench49.31117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.83116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.70115 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.89109 on the scale
- plot_unscramblingLiveBench79.05106 on the scale
- integrals_with_gameLiveBench100104 on the scale
- consecutive_eventsLiveBench90.21102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench70.5292 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench84.7887 on the scale
- paraphraseLiveBench71.5286 on the scale
- summarizeLiveBench69.4385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench83.1076 on the scale
- story_generationLiveBench75.9274 on the scale
- olympiadLiveBench91.6972 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench10041 on the scale
016openai/gpt-5.5-proxhigh122.6631
3 columns on 1 board, hardest first are the whole evidence behind 122.66. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.62119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.78116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.88109 on the scale
017anthropic/claude-fable-5max122.58352
35 columns on 2 boards, hardest first are the whole evidence behind 122.58. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.40123 on the scale
- typescriptLiveBench53.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.52121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.36120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.47119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.70119 on the scale
- tablejoinLiveBench56.13117 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub1116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.90116 on the scale
- pythonLiveBench65109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.87109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub1106 on the scale
- plot_unscramblingLiveBench78.72106 on the scale
- integrals_with_gameLiveBench97104 on the scale
- consecutive_eventsLiveBench91.36102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7896 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub1.0094 on the scale
- simplifyLiveBench72.0392 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench77.6586 on the scale
- summarizeLiveBench81.7085 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
- typosLiveBench9477 on the scale
- code_generationLiveBench91.5576 on the scale
- story_generationLiveBench71.7074 on the scale
- olympiadLiveBench92.8472 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench94.1241 on the scale
018anthropic/claude-opus-5max122.24342
34 columns on 2 boards, hardest first are the whole evidence behind 122.24. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.46123 on the scale
- typescriptLiveBench43.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.59121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.61120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.43119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.88119 on the scale
- tablejoinLiveBench51.96117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.73116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.60115 on the scale
- pythonLiveBench75109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.86109 on the scale
- plot_unscramblingLiveBench74.73106 on the scale
- integrals_with_gameLiveBench97104 on the scale
- consecutive_eventsLiveBench77.57102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench8696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench61.5892 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench65.7386 on the scale
- summarizeLiveBench66.4385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
- typosLiveBench9277 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench61.3274 on the scale
- olympiadLiveBench92.8072 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench99.0147 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench94.1241 on the scale
019claude-sonnet-5-5not in the catalogmax122.13362
36 columns on 2 boards, hardest first are the whole evidence behind 122.13. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0.03153 on the scale
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0.01137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0.03136 on the scale
- typescriptLiveBench56.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.65121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.75120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.59119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.98119 on the scale
- tablejoinLiveBench39.81117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.80116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.47115 on the scale
- pythonLiveBench35109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.89109 on the scale
- plot_unscramblingLiveBench48.62106 on the scale
- integrals_with_gameLiveBench97104 on the scale
- consecutive_eventsLiveBench40.62102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench8296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- simplifyLiveBench50.8592 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench86.9687 on the scale
- paraphraseLiveBench48.2286 on the scale
- summarizeLiveBench58.1785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9683 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench95.7876 on the scale
- story_generationLiveBench69.9274 on the scale
- olympiadLiveBench92.4772 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
020muse-wonder-1.0not in the catalogmax120.86302
30 columns on 2 boards, hardest first are the whole evidence behind 120.86. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.31123 on the scale
- typescriptLiveBench60122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.51121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.54120 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.46119 on the scale
- tablejoinLiveBench46.33117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.71116 on the scale
- pythonLiveBench60109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.82109 on the scale
- plot_unscramblingLiveBench74.80106 on the scale
- integrals_with_gameLiveBench98104 on the scale
- consecutive_eventsLiveBench98.53102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench8096 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench67.0592 on the scale
- connectionsLiveBench9691 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench74.0786 on the scale
- summarizeLiveBench76.8385 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench83.1076 on the scale
- story_generationLiveBench74.1774 on the scale
- olympiadLiveBench91.8272 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
021openai/gpt-5.5xhigh120.29322
32 columns on 2 boards, hardest first are the whole evidence behind 120.29. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.34123 on the scale
- typescriptLiveBench43.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.56121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.44120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.63119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.23119 on the scale
- tablejoinLiveBench55.90117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.72116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.63115 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.85109 on the scale
- plot_unscramblingLiveBench74.09106 on the scale
- integrals_with_gameLiveBench99104 on the scale
- consecutive_eventsLiveBench88.82102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7696 on the scale
- simplifyLiveBench66.3292 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench71.7386 on the scale
- summarizeLiveBench72.3285 on the scale
- typosLiveBench8877 on the scale
- code_generationLiveBench81.6976 on the scale
- story_generationLiveBench72.5574 on the scale
- olympiadLiveBench91.3572 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench10041 on the scale
022claude-sonnet-5-5not in the catalogxhigh119.88231
23 columns on 1 board, hardest first are the whole evidence behind 119.88. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- tablejoinLiveBench46.88117 on the scale
- pythonLiveBench35109 on the scale
- plot_unscramblingLiveBench67.79106 on the scale
- integrals_with_gameLiveBench100104 on the scale
- consecutive_eventsLiveBench90.94102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench36.3699 on the scale
- logic_with_navigationLiveBench8096 on the scale
- simplifyLiveBench72.3792 on the scale
- connectionsLiveBench98.5091 on the scale
- theory_of_mindLiveBench69.2388 on the scale
- code_completionLiveBench84.7887 on the scale
- paraphraseLiveBench62.2886 on the scale
- summarizeLiveBench70.3385 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench92.9676 on the scale
- story_generationLiveBench77.0874 on the scale
- olympiadLiveBench91.7272 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
023anthropic/claude-fable-5high119.2951
5 columns on 1 board, hardest first are the whole evidence behind 119.29. Each is a door to the column and everything else scored on it.
- OEIS Open LiteEpoch AI Benchmarking Hub0.44122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.45119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.64119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
024openai/gpt-5.6-terramax119.24322
32 columns on 2 boards, hardest first are the whole evidence behind 119.24. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.35121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.54120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.47119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.41119 on the scale
- tablejoinLiveBench50.25117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.71116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.43115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.86109 on the scale
- plot_unscramblingLiveBench69.68106 on the scale
- integrals_with_gameLiveBench95104 on the scale
- consecutive_eventsLiveBench87.67102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub1.0094 on the scale
- simplifyLiveBench61.4392 on the scale
- connectionsLiveBench9391 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench57.0386 on the scale
- summarizeLiveBench73.1785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench66.8374 on the scale
- olympiadLiveBench91.5372 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench10041 on the scale
025openai/gpt-5.4xhigh119.02231
23 columns on 1 board, hardest first are the whole evidence behind 119.02. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- tablejoinLiveBench51.75117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench65.90106 on the scale
- integrals_with_gameLiveBench93104 on the scale
- consecutive_eventsLiveBench86.19102 on the scale
- zebra_puzzleLiveBench98102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench7092 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench88.4688 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench63.4386 on the scale
- summarizeLiveBench68.4285 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench74.6576 on the scale
- story_generationLiveBench79.0274 on the scale
- olympiadLiveBench91.4772 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench10041 on the scale
026kimi-k3.1-pro-togethernot in the catalogmax118.88131
13 columns on 1 board, hardest first are the whole evidence behind 118.88. Each is a door to the column and everything else scored on it.
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0136 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.37123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.42121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.56120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.45119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.43119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.66116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.62115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.78109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
027deepseek/deepseek-v4-pro-0813unstated118.60231
23 columns on 1 board, hardest first are the whole evidence behind 118.60. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- tablejoinLiveBench49.46117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench64.22106 on the scale
- integrals_with_gameLiveBench94104 on the scale
- consecutive_eventsLiveBench90.23102 on the scale
- zebra_puzzleLiveBench96.75102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6696 on the scale
- simplifyLiveBench61.5792 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench68.6386 on the scale
- summarizeLiveBench65.6785 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench74.9374 on the scale
- olympiadLiveBench91.2972 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
028qwen3.8-maxnot in the catalogunstated118.08231
23 columns on 1 board, hardest first are the whole evidence behind 118.08. Each is a door to the column and everything else scored on it.
- typescriptLiveBench56.67122 on the scale
- tablejoinLiveBench50.06117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench58.56106 on the scale
- integrals_with_gameLiveBench81104 on the scale
- consecutive_eventsLiveBench87.14102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench7496 on the scale
- simplifyLiveBench67.3592 on the scale
- connectionsLiveBench94.5091 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench73.9187 on the scale
- paraphraseLiveBench73.3286 on the scale
- summarizeLiveBench79.1585 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench71.8376 on the scale
- story_generationLiveBench76.5274 on the scale
- olympiadLiveBench91.1572 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
029gpt-5.4-pro-2026-03-05not in the catalogxhigh117.9261
6 columns on 1 board, hardest first are the whole evidence behind 117.92. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.59119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.59116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.46115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.50115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.82109 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9583 on the scale
030x-ai/grok-4.6unstated117.62231
23 columns on 1 board, hardest first are the whole evidence behind 117.62. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- tablejoinLiveBench43.42117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench67.09106 on the scale
- integrals_with_gameLiveBench85104 on the scale
- consecutive_eventsLiveBench78.15102 on the scale
- zebra_puzzleLiveBench95.50102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench8296 on the scale
- simplifyLiveBench69.3392 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench71.7886 on the scale
- summarizeLiveBench68.0885 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench78.2774 on the scale
- olympiadLiveBench91.2172 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
031gpt-5.5-pro-pre-releasenot in the catalogxhigh117.3841
4 columns on 1 board, hardest first are the whole evidence behind 117.38. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.40120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.51115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
032meta/muse-spark-1.2xhigh117.04252
25 columns on 2 boards, hardest first are the whole evidence behind 117.04. Each is a door to the column and everything else scored on it.
- typescriptLiveBench50122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- tablejoinLiveBench46.83117 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.60115 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench61.72106 on the scale
- integrals_with_gameLiveBench91104 on the scale
- consecutive_eventsLiveBench82.55102 on the scale
- zebra_puzzleLiveBench95.25102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench8896 on the scale
- simplifyLiveBench68.7392 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench74.3086 on the scale
- summarizeLiveBench80.6285 on the scale
- typosLiveBench7477 on the scale
- code_generationLiveBench74.6576 on the scale
- story_generationLiveBench73.6574 on the scale
- olympiadLiveBench87.7572 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench8947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
033claude-opus-4-8not in the catalogmax116.89352
35 columns on 2 boards, hardest first are the whole evidence behind 116.89. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.29123 on the scale
- typescriptLiveBench33.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.36121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.31120 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.42120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.52119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.66119 on the scale
- tablejoinLiveBench51.08117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.56116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.53115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.47115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.80109 on the scale
- plot_unscramblingLiveBench61.65106 on the scale
- integrals_with_gameLiveBench89104 on the scale
- consecutive_eventsLiveBench52.91102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7896 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- simplifyLiveBench6292 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench84.7887 on the scale
- paraphraseLiveBench67.8786 on the scale
- summarizeLiveBench77.6785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench80.6074 on the scale
- olympiadLiveBench92.2372 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench98.0442 on the scale
- tablereformatLiveBench94.1241 on the scale
034gpt-5.5-pre-releasenot in the catalogxhigh116.8651
5 columns on 1 board, hardest first are the whole evidence behind 116.86. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.35120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.52115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.8184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
035meta/muse-spark-1.3xhigh116.79282
28 columns on 2 boards, hardest first are the whole evidence behind 116.79. Each is a door to the column and everything else scored on it.
- typescriptLiveBench60122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.14121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.34119 on the scale
- tablejoinLiveBench48.19117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.41116 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.74109 on the scale
- plot_unscramblingLiveBench68.38106 on the scale
- integrals_with_gameLiveBench96104 on the scale
- consecutive_eventsLiveBench90.50102 on the scale
- zebra_puzzleLiveBench98102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench7696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench72.5292 on the scale
- connectionsLiveBench9491 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench78.6286 on the scale
- summarizeLiveBench83.4085 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench81.6976 on the scale
- story_generationLiveBench77.4874 on the scale
- olympiadLiveBench91.7472 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
036gpt-5.2-pro-2025-12-11not in the catalogxhigh116.7431
3 columns on 1 board, hardest first are the whole evidence behind 116.74. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.60119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.46116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.74109 on the scale
037smaug-flashnot in the catalogunstated116.63231
23 columns on 1 board, hardest first are the whole evidence behind 116.63. Each is a door to the column and everything else scored on it.
- typescriptLiveBench50122 on the scale
- tablejoinLiveBench50.71117 on the scale
- pythonLiveBench65109 on the scale
- plot_unscramblingLiveBench60.31106 on the scale
- integrals_with_gameLiveBench76104 on the scale
- consecutive_eventsLiveBench88.61102 on the scale
- zebra_puzzleLiveBench98102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7296 on the scale
- simplifyLiveBench63.1392 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench73.9187 on the scale
- paraphraseLiveBench65.5786 on the scale
- summarizeLiveBench78.3585 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench69.9874 on the scale
- olympiadLiveBench90.2072 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
038moonshotai/kimi-k3unstated116.54231
23 columns on 1 board, hardest first are the whole evidence behind 116.54. Each is a door to the column and everything else scored on it.
- typescriptLiveBench53.33122 on the scale
- tablejoinLiveBench48.40117 on the scale
- pythonLiveBench65109 on the scale
- plot_unscramblingLiveBench72.58106 on the scale
- integrals_with_gameLiveBench54104 on the scale
- consecutive_eventsLiveBench89.76102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench8096 on the scale
- simplifyLiveBench66.3792 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench82.6988 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench74.3786 on the scale
- summarizeLiveBench69.3585 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench75.3774 on the scale
- olympiadLiveBench91.6572 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
039deepseek/deepseek-v4.1-flashmax116.49322
32 columns on 2 boards, hardest first are the whole evidence behind 116.49. Each is a door to the column and everything else scored on it.
- typescriptLiveBench60122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.43121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.34120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.48119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.64119 on the scale
- tablejoinLiveBench49.04117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.43115 on the scale
- pythonLiveBench90109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.67109 on the scale
- plot_unscramblingLiveBench61.57106 on the scale
- integrals_with_gameLiveBench90104 on the scale
- consecutive_eventsLiveBench88.72102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench81.8299 on the scale
- logic_with_navigationLiveBench6896 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- simplifyLiveBench67.0792 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench82.6187 on the scale
- paraphraseLiveBench70.7386 on the scale
- summarizeLiveBench69.6585 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench72.6874 on the scale
- olympiadLiveBench89.0872 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench10041 on the scale
040qwen3.8-maxnot in the catalogxhigh116.3981
8 columns on 1 board, hardest first are the whole evidence behind 116.39. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.38121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.46119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.52119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.46116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.46115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.75109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
041deepseek/deepseek-v4-flash-vision-expunstated116.36231
23 columns on 1 board, hardest first are the whole evidence behind 116.36. Each is a door to the column and everything else scored on it.
- typescriptLiveBench56.67122 on the scale
- tablejoinLiveBench50.35117 on the scale
- pythonLiveBench75109 on the scale
- plot_unscramblingLiveBench57.75106 on the scale
- integrals_with_gameLiveBench76104 on the scale
- consecutive_eventsLiveBench90.05102 on the scale
- zebra_puzzleLiveBench98.75102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench65.0792 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench67.3987 on the scale
- paraphraseLiveBench72.8386 on the scale
- summarizeLiveBench72.1085 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench69.0176 on the scale
- story_generationLiveBench73.8374 on the scale
- olympiadLiveBench90.1772 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench8847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
042smaug-agenticnot in the catalogunstated116.30231
23 columns on 1 board, hardest first are the whole evidence behind 116.30. Each is a door to the column and everything else scored on it.
- typescriptLiveBench56.67122 on the scale
- tablejoinLiveBench51.23117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench71.07106 on the scale
- integrals_with_gameLiveBench52104 on the scale
- consecutive_eventsLiveBench90.42102 on the scale
- zebra_puzzleLiveBench98.25102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench8496 on the scale
- simplifyLiveBench66.6292 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench68.6386 on the scale
- summarizeLiveBench72.0585 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench84.5176 on the scale
- story_generationLiveBench76.6774 on the scale
- olympiadLiveBench91.5672 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench98.0441 on the scale
043deepseek/deepseek-v4.1-flash-1017max116.2571
7 columns on 1 board, hardest first are the whole evidence behind 116.25. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.49121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.46119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.46116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.42115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.71109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
044gpt-5.4-2026-03-05not in the catalogxhigh116.23131
13 columns on 1 board, hardest first are the whole evidence behind 116.23. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.25123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.37121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.27120 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.38120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.48119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.23119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0.50116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.49116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.45115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.48115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.79109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
045google/gemini-3.7-flashhigh116.01322
32 columns on 2 boards, hardest first are the whole evidence behind 116.01. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.37121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.27120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.66119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.41119 on the scale
- tablejoinLiveBench45.48117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.37116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.69115 on the scale
- pythonLiveBench60109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.72109 on the scale
- plot_unscramblingLiveBench66.36106 on the scale
- integrals_with_gameLiveBench88104 on the scale
- consecutive_eventsLiveBench60.37102 on the scale
- zebra_puzzleLiveBench94.50102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9794 on the scale
- simplifyLiveBench74.6292 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench82.6988 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench78.0786 on the scale
- summarizeLiveBench84.7785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9583 on the scale
- typosLiveBench9077 on the scale
- code_generationLiveBench81.6976 on the scale
- story_generationLiveBench82.2574 on the scale
- olympiadLiveBench91.7972 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench98.0441 on the scale
046gpt-5.4-2026-03-05not in the cataloghigh115.9871
7 columns on 1 board, hardest first are the whole evidence behind 115.98. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.42119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.16119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0.50116 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7784 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
047anthropic/claude-opus-5unstated115.9431
3 columns on 1 board, hardest first are the whole evidence behind 115.94. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.37121 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
048meta/muse-spark-1.3max115.5551
5 columns on 1 board, hardest first are the whole evidence behind 115.55. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.25121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.54119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.46116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.74109 on the scale
049openai/gpt-5.6-lunamax114.95322
32 columns on 2 boards, hardest first are the whole evidence behind 114.95. Each is a door to the column and everything else scored on it.
- typescriptLiveBench36.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.21121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.42120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.43119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.21119 on the scale
- tablejoinLiveBench47.60117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.61116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.41115 on the scale
- pythonLiveBench45109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.82109 on the scale
- plot_unscramblingLiveBench51.20106 on the scale
- integrals_with_gameLiveBench70104 on the scale
- consecutive_eventsLiveBench86.50102 on the scale
- zebra_puzzleLiveBench93.50102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8096 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- simplifyLiveBench58.6792 on the scale
- connectionsLiveBench96.5091 on the scale
- theory_of_mindLiveBench73.0888 on the scale
- code_completionLiveBench86.9687 on the scale
- paraphraseLiveBench47.8286 on the scale
- summarizeLiveBench64.9385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9283 on the scale
- typosLiveBench7077 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench69.0774 on the scale
- olympiadLiveBench88.6572 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench92.1642 on the scale
- tablereformatLiveBench10041 on the scale
050doubao-seed-evolvingnot in the cataloghigh114.7641
4 columns on 1 board, hardest first are the whole evidence behind 114.76. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.40119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.42115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
051moonshotai/kimi-k3max114.75101
10 columns on 1 board, hardest first are the whole evidence behind 114.75. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.26121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.34120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.54119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.39116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.51115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.72109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
052x-ai/grok-4.5unstated114.74231
23 columns on 1 board, hardest first are the whole evidence behind 114.74. Each is a door to the column and everything else scored on it.
- typescriptLiveBench36.67122 on the scale
- tablejoinLiveBench43.58117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench66.39106 on the scale
- integrals_with_gameLiveBench78104 on the scale
- consecutive_eventsLiveBench75.54102 on the scale
- zebra_puzzleLiveBench94102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench7296 on the scale
- simplifyLiveBench70.1792 on the scale
- connectionsLiveBench9891 on the scale
- theory_of_mindLiveBench82.6988 on the scale
- code_completionLiveBench69.5687 on the scale
- paraphraseLiveBench69.2286 on the scale
- summarizeLiveBench73.1585 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench67.6176 on the scale
- story_generationLiveBench73.5874 on the scale
- olympiadLiveBench89.2472 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
053smaug-mininot in the catalogunstated114.73231
23 columns on 1 board, hardest first are the whole evidence behind 114.73. Each is a door to the column and everything else scored on it.
- typescriptLiveBench53.33122 on the scale
- tablejoinLiveBench51.87117 on the scale
- pythonLiveBench70109 on the scale
- plot_unscramblingLiveBench55.33106 on the scale
- integrals_with_gameLiveBench76104 on the scale
- consecutive_eventsLiveBench88.50102 on the scale
- zebra_puzzleLiveBench94102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench65.3892 on the scale
- connectionsLiveBench91.6791 on the scale
- theory_of_mindLiveBench69.2388 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench76.1886 on the scale
- summarizeLiveBench76.1085 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench70.4276 on the scale
- story_generationLiveBench7874 on the scale
- olympiadLiveBench90.8572 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench96.0841 on the scale
054x-ai/grok-4.6xhigh114.39101
10 columns on 1 board, hardest first are the whole evidence behind 114.39. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.30123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.34121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.40120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.33119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.41119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.32116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.49115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.66109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
056google/gemini-3.8-flashhigh114.21332
33 columns on 2 boards, hardest first are the whole evidence behind 114.21. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- typescriptLiveBench40122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.47121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.32120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.58119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.39119 on the scale
- tablejoinLiveBench45.33117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.22116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.70115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.68109 on the scale
- plot_unscramblingLiveBench73.38106 on the scale
- integrals_with_gameLiveBench80104 on the scale
- consecutive_eventsLiveBench20.61102 on the scale
- zebra_puzzleLiveBench98.25102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench8296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench75.4092 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench71.7487 on the scale
- paraphraseLiveBench8086 on the scale
- summarizeLiveBench87.5085 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9583 on the scale
- typosLiveBench9077 on the scale
- code_generationLiveBench73.2476 on the scale
- story_generationLiveBench82.7574 on the scale
- olympiadLiveBench92.1672 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench96.0841 on the scale
055meta/muse-spark-1.1xhigh114.21242
24 columns on 2 boards, hardest first are the whole evidence behind 114.21. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.34119 on the scale
- tablejoinLiveBench46.56117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench63.03106 on the scale
- integrals_with_gameLiveBench83104 on the scale
- consecutive_eventsLiveBench71.09102 on the scale
- zebra_puzzleLiveBench96102 on the scale
- javascriptLiveBench77.2799 on the scale
- logic_with_navigationLiveBench7896 on the scale
- simplifyLiveBench65.7092 on the scale
- connectionsLiveBench9891 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench69.4086 on the scale
- summarizeLiveBench74.3085 on the scale
- typosLiveBench6277 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench69.1574 on the scale
- olympiadLiveBench87.5072 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench8247 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench10041 on the scale
057claude-sonnet-4-6-thinking-autonot in the catalogmedium114.06231
23 columns on 1 board, hardest first are the whole evidence behind 114.06. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- tablejoinLiveBench45.08117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench56.98106 on the scale
- integrals_with_gameLiveBench90104 on the scale
- consecutive_eventsLiveBench92.68102 on the scale
- zebra_puzzleLiveBench96102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench62.9392 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench73.0888 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench59.4086 on the scale
- summarizeLiveBench61.6585 on the scale
- typosLiveBench7277 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench68.9074 on the scale
- olympiadLiveBench87.8672 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench7647 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench96.0841 on the scale
058x-ai/grok-4.6high114.0241
4 columns on 1 board, hardest first are the whole evidence behind 114.02. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.33119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.49115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
059openai/gpt-6-lunamax113.99352
35 columns on 2 boards, hardest first are the whole evidence behind 113.99. Each is a door to the column and everything else scored on it.
- SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub0139 on the scale
- SpireClimbEpoch AI Benchmarking Hub0137 on the scale
- SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub0136 on the scale
- typescriptLiveBench40122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.07121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.44120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.27119 on the scale
- tablejoinLiveBench49.62117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.56116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.41115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.79109 on the scale
- plot_unscramblingLiveBench57.48106 on the scale
- integrals_with_gameLiveBench74104 on the scale
- consecutive_eventsLiveBench70.49102 on the scale
- zebra_puzzleLiveBench96102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench55.4392 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench73.0888 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench49.7086 on the scale
- summarizeLiveBench55.3285 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
- typosLiveBench6477 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench63.2874 on the scale
- olympiadLiveBench90.3872 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench10041 on the scale
060muse-spark-open-816b-a42bnot in the catalogxhigh113.9251
5 columns on 1 board, hardest first are the whole evidence behind 113.92. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.32116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.57115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.70109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
061google/gemini-3.1-pro-previewhigh113.89302
30 columns on 2 boards, hardest first are the whole evidence behind 113.89. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.34121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.27120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.51119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.09119 on the scale
- tablejoinLiveBench52.35117 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.73115 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench74.13106 on the scale
- integrals_with_gameLiveBench78104 on the scale
- consecutive_eventsLiveBench85.24102 on the scale
- zebra_puzzleLiveBench85.25102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- simplifyLiveBench71.0592 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench79.8786 on the scale
- summarizeLiveBench88.2385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench74.6576 on the scale
- story_generationLiveBench77.2574 on the scale
- olympiadLiveBench92.1072 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench98.0441 on the scale
062claude-haiku-5-5not in the catalogmax113.88322
32 columns on 2 boards, hardest first are the whole evidence behind 113.88. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.44123 on the scale
- typescriptLiveBench43.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.30121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.47120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.55119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.79119 on the scale
- tablejoinLiveBench46.67117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.46116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.24115 on the scale
- pythonLiveBench60109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.75109 on the scale
- plot_unscramblingLiveBench27.88106 on the scale
- integrals_with_gameLiveBench67104 on the scale
- consecutive_eventsLiveBench10.64102 on the scale
- zebra_puzzleLiveBench94102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6696 on the scale
- simplifyLiveBench50.2592 on the scale
- connectionsLiveBench9691 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench50.2286 on the scale
- summarizeLiveBench53.6385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
- typosLiveBench6477 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench64.5374 on the scale
- olympiadLiveBench83.7072 on the scale
- spatialLiveBench9253 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench98.0442 on the scale
- tablereformatLiveBench98.0441 on the scale
063google/gemini-3-flash-previewhigh113.7051
5 columns on 1 board, hardest first are the whole evidence behind 113.70. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.20121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.40119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.67115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
064deepseek/deepseek-v4-pro-0813max113.6081
8 columns on 1 board, hardest first are the whole evidence behind 113.60. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.43121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.46119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.33119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.53115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.65109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9283 on the scale
065qwen/qwen3.8-max-0902xhigh113.5181
8 columns on 1 board, hardest first are the whole evidence behind 113.51. Each is a door to the column and everything else scored on it.
- Furniture AssemblyEpoch AI Benchmarking Hub0.20120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.48119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.34116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.47115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.66109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9283 on the scale
066anthropic/claude-sonnet-5xhigh113.50272
27 columns on 2 boards, hardest first are the whole evidence behind 113.50. Each is a door to the column and everything else scored on it.
- typescriptLiveBench50122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.34119 on the scale
- tablejoinLiveBench42.44117 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.33115 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench53.41106 on the scale
- integrals_with_gameLiveBench88104 on the scale
- consecutive_eventsLiveBench74.74102 on the scale
- zebra_puzzleLiveBench88102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench8696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9594 on the scale
- simplifyLiveBench60.6592 on the scale
- connectionsLiveBench95.5091 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench61.2586 on the scale
- summarizeLiveBench71.2285 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
- typosLiveBench7677 on the scale
- code_generationLiveBench83.1076 on the scale
- story_generationLiveBench62.3274 on the scale
- olympiadLiveBench90.6772 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
067deepseek/deepseek-v4-flash-0731unstated113.47231
23 columns on 1 board, hardest first are the whole evidence behind 113.47. Each is a door to the column and everything else scored on it.
- typescriptLiveBench26.67122 on the scale
- tablejoinLiveBench48.54117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench58.19106 on the scale
- integrals_with_gameLiveBench65104 on the scale
- consecutive_eventsLiveBench89.44102 on the scale
- zebra_puzzleLiveBench100102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench58.3792 on the scale
- connectionsLiveBench97.3391 on the scale
- theory_of_mindLiveBench86.5488 on the scale
- code_completionLiveBench73.9187 on the scale
- paraphraseLiveBench61.7086 on the scale
- summarizeLiveBench71.5085 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench70.5074 on the scale
- olympiadLiveBench89.0872 on the scale
- spatialLiveBench9053 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench10041 on the scale
068gpt-5.2-2025-12-11not in the catalogxhigh113.39111
11 columns on 1 board, hardest first are the whole evidence behind 113.39. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.23123 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.19120 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.38120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.48119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.32119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.32116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.37115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.41115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.67109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
069x-ai/grok-4.7xhigh113.32332
33 columns on 2 boards, hardest first are the whole evidence behind 113.32. Each is a door to the column and everything else scored on it.
- FrontierMath-ErdosEpoch AI Benchmarking Hub0153 on the scale
- typescriptLiveBench43.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.29121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.21120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.41119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.54119 on the scale
- tablejoinLiveBench44.92117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.17116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.56115 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.53109 on the scale
- plot_unscramblingLiveBench72.42106 on the scale
- integrals_with_gameLiveBench96104 on the scale
- consecutive_eventsLiveBench85.73102 on the scale
- zebra_puzzleLiveBench86102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- simplifyLiveBench69.9792 on the scale
- connectionsLiveBench9891 on the scale
- theory_of_mindLiveBench84.6188 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench78.0586 on the scale
- summarizeLiveBench78.1785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
- typosLiveBench7077 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench74.9274 on the scale
- olympiadLiveBench91.6572 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
070claude-opus-4-7not in the catalogxhigh113.14312
31 columns on 2 boards, hardest first are the whole evidence behind 113.14. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.23120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- tablejoinLiveBench47.23117 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.52115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.44115 on the scale
- pythonLiveBench45109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.90106 on the scale
- plot_unscramblingLiveBench61.41106 on the scale
- integrals_with_gameLiveBench84104 on the scale
- consecutive_eventsLiveBench89.52102 on the scale
- zebra_puzzleLiveBench98102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7096 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- simplifyLiveBench65.5092 on the scale
- connectionsLiveBench92.3391 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench65.7786 on the scale
- summarizeLiveBench66.2885 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
- typosLiveBench8077 on the scale
- code_generationLiveBench85.9276 on the scale
- story_generationLiveBench69.4074 on the scale
- olympiadLiveBench91.3872 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench98.0442 on the scale
- tablereformatLiveBench98.0441 on the scale
071qwen3.8-flash-nextnot in the catalogunstated113.12231
23 columns on 1 board, hardest first are the whole evidence behind 113.12. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- tablejoinLiveBench51.02117 on the scale
- pythonLiveBench70109 on the scale
- plot_unscramblingLiveBench55.76106 on the scale
- integrals_with_gameLiveBench59104 on the scale
- consecutive_eventsLiveBench75.62102 on the scale
- zebra_puzzleLiveBench98.75102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench70.2592 on the scale
- connectionsLiveBench90.1791 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench79.4786 on the scale
- summarizeLiveBench77.3585 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench69.0176 on the scale
- story_generationLiveBench81.3774 on the scale
- olympiadLiveBench91.1772 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench96.0841 on the scale
072claude-opus-4-6-thinking-autonot in the cataloghigh113.04231
23 columns on 1 board, hardest first are the whole evidence behind 113.04. Each is a door to the column and everything else scored on it.
- typescriptLiveBench33.33122 on the scale
- tablejoinLiveBench48.17117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench66.48106 on the scale
- integrals_with_gameLiveBench73104 on the scale
- consecutive_eventsLiveBench63.47102 on the scale
- zebra_puzzleLiveBench94102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8096 on the scale
- simplifyLiveBench59.1592 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench82.6988 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench62.8286 on the scale
- summarizeLiveBench65.1585 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench66.1374 on the scale
- olympiadLiveBench92.1772 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
073claude-opus-4-7not in the catalogmax113.03101
10 columns on 1 board, hardest first are the whole evidence behind 113.03. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.19123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.28121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.33120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.41119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.43119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.32116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.70109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.8384 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
074z-ai/glm-5.3unstated112.92231
23 columns on 1 board, hardest first are the whole evidence behind 112.92. Each is a door to the column and everything else scored on it.
- typescriptLiveBench50122 on the scale
- tablejoinLiveBench42.35117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench60.24106 on the scale
- integrals_with_gameLiveBench69104 on the scale
- consecutive_eventsLiveBench72.31102 on the scale
- zebra_puzzleLiveBench88.75102 on the scale
- javascriptLiveBench72.7399 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench65.8592 on the scale
- connectionsLiveBench95.3391 on the scale
- theory_of_mindLiveBench88.4688 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench64.6886 on the scale
- summarizeLiveBench74.5585 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench72.1374 on the scale
- olympiadLiveBench87.5172 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench96.0841 on the scale
075claude-haiku-5-5not in the catalogxhigh112.88242
24 columns on 2 boards, hardest first are the whole evidence behind 112.88. Each is a door to the column and everything else scored on it.
- typescriptLiveBench40122 on the scale
- tablejoinLiveBench49.17117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench42.72106 on the scale
- integrals_with_gameLiveBench89104 on the scale
- consecutive_eventsLiveBench71.05102 on the scale
- zebra_puzzleLiveBench98102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench5696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- simplifyLiveBench62.7892 on the scale
- connectionsLiveBench93.8391 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench73.9187 on the scale
- paraphraseLiveBench62.5086 on the scale
- summarizeLiveBench66.9885 on the scale
- typosLiveBench5477 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench73.6074 on the scale
- olympiadLiveBench86.9572 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
076qwen/qwen3.8-27bunstated112.72231
23 columns on 1 board, hardest first are the whole evidence behind 112.72. Each is a door to the column and everything else scored on it.
- typescriptLiveBench50122 on the scale
- tablejoinLiveBench48.75117 on the scale
- pythonLiveBench75109 on the scale
- plot_unscramblingLiveBench56.54106 on the scale
- integrals_with_gameLiveBench65104 on the scale
- consecutive_eventsLiveBench84.93102 on the scale
- zebra_puzzleLiveBench96.50102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench66.7092 on the scale
- connectionsLiveBench90.5091 on the scale
- theory_of_mindLiveBench59.6288 on the scale
- code_completionLiveBench73.9187 on the scale
- paraphraseLiveBench74.9086 on the scale
- summarizeLiveBench75.9385 on the scale
- typosLiveBench7677 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench73.1074 on the scale
- olympiadLiveBench87.7372 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench96.0841 on the scale
077z-ai/glm-5.3max112.5471
7 columns on 1 board, hardest first are the whole evidence behind 112.54. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.33121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.29116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.41115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.69109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
078anthropic/claude-sonnet-5max112.4381
8 columns on 1 board, hardest first are the whole evidence behind 112.43. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.35121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.45119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.63119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.29116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.34115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.66109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8083 on the scale
079moonshotai/kimi-k3high112.2031
3 columns on 1 board, hardest first are the whole evidence behind 112.20. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9283 on the scale
080gpt-5.4-2026-03-05not in the catalogmedium112.0441
4 columns on 1 board, hardest first are the whole evidence behind 112.04. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.28121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
081claude-opus-4-6-64knot in the catalogunstated111.9261
6 columns on 1 board, hardest first are the whole evidence behind 111.92. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.21120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.40115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.90106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
082claude-opus-4-6not in the catalogmax111.69141
14 columns on 1 board, hardest first are the whole evidence behind 111.69. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.13123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.25121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.23120 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.28120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.33119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.47115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.41115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.66109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.90106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
083google/gemini-3.1-pro-previewunstated111.6691
9 columns on 1 board, hardest first are the whole evidence behind 111.66. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.14123 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.17120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.37115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.60109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.89106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
084deepseek/deepseek-v4-prounstated111.64231
23 columns on 1 board, hardest first are the whole evidence behind 111.64. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- tablejoinLiveBench48.19117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench56.39106 on the scale
- integrals_with_gameLiveBench78104 on the scale
- consecutive_eventsLiveBench75.42102 on the scale
- zebra_puzzleLiveBench92102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench6496 on the scale
- simplifyLiveBench55.9392 on the scale
- connectionsLiveBench9891 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench69.5687 on the scale
- paraphraseLiveBench56.2286 on the scale
- summarizeLiveBench69.5085 on the scale
- typosLiveBench8077 on the scale
- code_generationLiveBench70.4276 on the scale
- story_generationLiveBench67.7574 on the scale
- olympiadLiveBench90.6272 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench10041 on the scale
085x-ai/grok-4.5high111.5581
8 columns on 1 board, hardest first are the whole evidence behind 111.55. Each is a door to the column and everything else scored on it.
- Furniture AssemblyEpoch AI Benchmarking Hub0.23120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.35119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.32119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.24116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.48115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.57109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
086claude-opus-4-8not in the cataloglow111.5431
3 columns on 1 board, hardest first are the whole evidence behind 111.54. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.29119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
087deepseek/deepseek-v4-flash-0731max111.1471
7 columns on 1 board, hardest first are the whole evidence behind 111.14. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.34121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.32119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.24116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.34115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.58109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
088google/gemini-3.5-flashhigh111.12382
38 columns on 2 boards, hardest first are the whole evidence behind 111.12. Each is a door to the column and everything else scored on it.
- OEIS OpenEpoch AI Benchmarking Hub0.22127 on the scale
- EBR-benchEpoch AI Benchmarking Hub0.05123 on the scale
- OEIS Open LiteEpoch AI Benchmarking Hub0.29122 on the scale
- typescriptLiveBench33.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.32121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.15120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.45119 on the scale
- tablejoinLiveBench50.50117 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.66115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.39115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.63109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
- plot_unscramblingLiveBench67.75106 on the scale
- integrals_with_gameLiveBench68104 on the scale
- consecutive_eventsLiveBench47.99102 on the scale
- zebra_puzzleLiveBench77.25102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- simplifyLiveBench69.1092 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench75.5286 on the scale
- summarizeLiveBench82.5885 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7984 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
- typosLiveBench8677 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench75.2074 on the scale
- olympiadLiveBench91.8872 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench96.0841 on the scale
089qwen/qwen3.7-maxunstated111.06322
32 columns on 2 boards, hardest first are the whole evidence behind 111.06. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.10123 on the scale
- typescriptLiveBench26.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.32121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- tablejoinLiveBench45.52117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.34116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.56115 on the scale
- pythonLiveBench45109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.65109 on the scale
- plot_unscramblingLiveBench58.72106 on the scale
- integrals_with_gameLiveBench59104 on the scale
- consecutive_eventsLiveBench71.80102 on the scale
- zebra_puzzleLiveBench74.50102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench8496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- simplifyLiveBench64.0892 on the scale
- connectionsLiveBench96.5091 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench69.5687 on the scale
- paraphraseLiveBench7286 on the scale
- summarizeLiveBench83.8085 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7784 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench76.2874 on the scale
- olympiadLiveBench86.9372 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
090claude-opus-4-6-32knot in the catalogunstated111.0361
6 columns on 1 board, hardest first are the whole evidence behind 111.03. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.21120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.40115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
091google/gemini-3.6-flashhigh110.95322
32 columns on 2 boards, hardest first are the whole evidence behind 110.95. Each is a door to the column and everything else scored on it.
- typescriptLiveBench26.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.30121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.23120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.40119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.23119 on the scale
- tablejoinLiveBench47.13117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.22116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.66115 on the scale
- pythonLiveBench40109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.59109 on the scale
- plot_unscramblingLiveBench63.69106 on the scale
- integrals_with_gameLiveBench63104 on the scale
- consecutive_eventsLiveBench43.82102 on the scale
- zebra_puzzleLiveBench85.75102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- simplifyLiveBench75.7092 on the scale
- connectionsLiveBench10091 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench74.3286 on the scale
- summarizeLiveBench75.8585 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9483 on the scale
- typosLiveBench8877 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench75.6074 on the scale
- olympiadLiveBench90.4972 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench98.0441 on the scale
092openai/gpt-5.6-sollow110.8541
4 columns on 1 board, hardest first are the whole evidence behind 110.85. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.26121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.31119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
093google/gemini-3.5-flashlow110.8341
4 columns on 1 board, hardest first are the whole evidence behind 110.83. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.28121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.42119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
094moonshotai/kimi-k2.6unstated110.82131
13 columns on 1 board, hardest first are the whole evidence behind 110.82. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.02123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.18121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.15120 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.22120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.20119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.26116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.35115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.39115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.57109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.90106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7784 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
095mistral-large-4-0-ossnot in the cataloghigh110.7871
7 columns on 1 board, hardest first are the whole evidence behind 110.78. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.18123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.20121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.31120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.29119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.22115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
097gemini-3-pro-previewnot in the catalogunstated110.7471
7 columns on 1 board, hardest first are the whole evidence behind 110.74. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.19120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.38115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7384 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9383 on the scale
096gpt-5.2-2025-12-11not in the cataloghigh110.74332
33 columns on 2 boards, hardest first are the whole evidence behind 110.74. Each is a door to the column and everything else scored on it.
- typescriptLiveBench36.67122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.23121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.19120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.45119 on the scale
- tablejoinLiveBench45.58117 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.34115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.40115 on the scale
- pythonLiveBench55109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
- plot_unscramblingLiveBench58.43106 on the scale
- integrals_with_gameLiveBench92104 on the scale
- consecutive_eventsLiveBench88.91102 on the scale
- zebra_puzzleLiveBench84102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- simplifyLiveBench54.0892 on the scale
- connectionsLiveBench9991 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench64.1086 on the scale
- summarizeLiveBench63.2085 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7484 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench65.7074 on the scale
- olympiadLiveBench88.5772 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench10041 on the scale
098claude-opus-4-5-20251101-thinking-64knot in the cataloghigh110.65231
23 columns on 1 board, hardest first are the whole evidence behind 110.65. Each is a door to the column and everything else scored on it.
- typescriptLiveBench20122 on the scale
- tablejoinLiveBench45.92117 on the scale
- pythonLiveBench40109 on the scale
- plot_unscramblingLiveBench66.45106 on the scale
- integrals_with_gameLiveBench78104 on the scale
- consecutive_eventsLiveBench79.36102 on the scale
- zebra_puzzleLiveBench77.50102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench54.9592 on the scale
- connectionsLiveBench99.3391 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench65.6786 on the scale
- summarizeLiveBench63.7285 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench65.8574 on the scale
- olympiadLiveBench89.4672 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
099mistralai/mistral-large-4-0high110.6391
9 columns on 1 board, hardest first are the whole evidence behind 110.63. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- Furniture AssemblyEpoch AI Benchmarking Hub0.21120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.33119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.40119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.27116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.20115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.59109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
100muse-sparknot in the catalogunstated110.4141
4 columns on 1 board, hardest first are the whole evidence behind 110.41. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.15120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.39115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
101z-ai/glm-5.2max110.3391
9 columns on 1 board, hardest first are the whole evidence behind 110.33. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.10123 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.22119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.37119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.29116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.34115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.59109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8694 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7984 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9283 on the scale
102openai/gpt-5.4-nanoxhigh110.30231
23 columns on 1 board, hardest first are the whole evidence behind 110.30. Each is a door to the column and everything else scored on it.
- typescriptLiveBench26.67122 on the scale
- tablejoinLiveBench50.59117 on the scale
- pythonLiveBench50109 on the scale
- plot_unscramblingLiveBench39.13106 on the scale
- integrals_with_gameLiveBench82.83104 on the scale
- consecutive_eventsLiveBench54.30102 on the scale
- zebra_puzzleLiveBench97.62102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6496 on the scale
- simplifyLiveBench62.7592 on the scale
- connectionsLiveBench94.3991 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench67.3987 on the scale
- paraphraseLiveBench66.6586 on the scale
- summarizeLiveBench71.7785 on the scale
- typosLiveBench5477 on the scale
- code_generationLiveBench74.2976 on the scale
- story_generationLiveBench67.6574 on the scale
- olympiadLiveBench87.9872 on the scale
- spatialLiveBench8253 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
103anthropic/claude-fable-5low110.2031
3 columns on 1 board, hardest first are the whole evidence behind 110.20. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.31119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7983 on the scale
104claude-opus-4-6not in the catalogunstated110.0841
4 columns on 1 board, hardest first are the whole evidence behind 110.08. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.15121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.15120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.38115 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7984 on the scale
105openai/gpt-5.2-codexunstated110.02231
23 columns on 1 board, hardest first are the whole evidence behind 110.02. Each is a door to the column and everything else scored on it.
- typescriptLiveBench20122 on the scale
- tablejoinLiveBench45.60117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench56.03106 on the scale
- integrals_with_gameLiveBench75104 on the scale
- consecutive_eventsLiveBench89.02102 on the scale
- zebra_puzzleLiveBench70102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench60.5792 on the scale
- connectionsLiveBench9591 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench86.9687 on the scale
- paraphraseLiveBench67.0586 on the scale
- summarizeLiveBench63.3885 on the scale
- typosLiveBench7077 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench74.7874 on the scale
- olympiadLiveBench87.0072 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench10041 on the scale
106claude-sonnet-4-6-16knot in the catalogunstated109.9841
4 columns on 1 board, hardest first are the whole evidence behind 109.98. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.08120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.32115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
107gpt-5.1-2025-11-13not in the cataloghigh109.9471
7 columns on 1 board, hardest first are the whole evidence behind 109.94. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.13120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.42119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.48115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.31115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.6884 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
108mistral-large-4not in the cataloghigh109.81231
23 columns on 1 board, hardest first are the whole evidence behind 109.81. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- tablejoinLiveBench49.10117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench40.93106 on the scale
- integrals_with_gameLiveBench90104 on the scale
- consecutive_eventsLiveBench80.53102 on the scale
- zebra_puzzleLiveBench93102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench66.1892 on the scale
- connectionsLiveBench4891 on the scale
- theory_of_mindLiveBench80.7788 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench63.9786 on the scale
- summarizeLiveBench61.6385 on the scale
- typosLiveBench6077 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench67.5074 on the scale
- olympiadLiveBench89.2872 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
109doubao-seed-2.1-pro-260628not in the cataloghigh109.6141
4 columns on 1 board, hardest first are the whole evidence behind 109.61. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.45115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
110z-ai/glm-5.2unstated109.40231
23 columns on 1 board, hardest first are the whole evidence behind 109.40. Each is a door to the column and everything else scored on it.
- typescriptLiveBench36.67122 on the scale
- tablejoinLiveBench45.87117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench60.73106 on the scale
- integrals_with_gameLiveBench76104 on the scale
- consecutive_eventsLiveBench79.28102 on the scale
- zebra_puzzleLiveBench65.50102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench8096 on the scale
- simplifyLiveBench55.6792 on the scale
- connectionsLiveBench9491 on the scale
- theory_of_mindLiveBench7588 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench62.2586 on the scale
- summarizeLiveBench68.3785 on the scale
- typosLiveBench7477 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench62.8874 on the scale
- olympiadLiveBench89.0572 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench96.0841 on the scale
111z-ai/glm-5.1unstated109.27101
10 columns on 1 board, hardest first are the whole evidence behind 109.27. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.13120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.20119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.34115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.33115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.37109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.80106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7484 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
112gpt-5-pro-2025-10-06not in the cataloghigh109.2351
5 columns on 1 board, hardest first are the whole evidence behind 109.23. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.15120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.48119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.20116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.56109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
113anthropic/claude-opus-5low109.1331
3 columns on 1 board, hardest first are the whole evidence behind 109.13. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
114doubao-seed-2.1-turbo-260628not in the cataloghigh108.8441
4 columns on 1 board, hardest first are the whole evidence behind 108.84. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.26119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.30115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
115gpt-5.2-2025-12-11not in the catalogmedium108.7591
9 columns on 1 board, hardest first are the whole evidence behind 108.75. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.22121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.17120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.33115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.37115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
116gpt-5-2025-08-07not in the cataloghigh108.61151
15 columns on 1 board, hardest first are the whole evidence behind 108.61. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.13123 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.23121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.13120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- MirrorCodeEpoch AI Benchmarking Hub0.04119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.22116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.50115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.32115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.55109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7484 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
117z-ai/glm-5.3-flashmax108.4561
6 columns on 1 board, hardest first are the whole evidence behind 108.45. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.08121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.17116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.56109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
118z-ai/glm-5.3-flashunstated108.42231
23 columns on 1 board, hardest first are the whole evidence behind 108.42. Each is a door to the column and everything else scored on it.
- typescriptLiveBench46.67122 on the scale
- tablejoinLiveBench48.83117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench60.26106 on the scale
- integrals_with_gameLiveBench45104 on the scale
- consecutive_eventsLiveBench84.30102 on the scale
- zebra_puzzleLiveBench79.50102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6296 on the scale
- simplifyLiveBench46.9592 on the scale
- connectionsLiveBench93.6791 on the scale
- theory_of_mindLiveBench73.0888 on the scale
- code_completionLiveBench80.4487 on the scale
- paraphraseLiveBench47.9886 on the scale
- summarizeLiveBench56.4785 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench77.4776 on the scale
- story_generationLiveBench59.8874 on the scale
- olympiadLiveBench89.8472 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench93.1442 on the scale
- tablereformatLiveBench96.0841 on the scale
119grok-4.20-0309-reasoningnot in the catalogunstated108.3361
6 columns on 1 board, hardest first are the whole evidence behind 108.33. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.29119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.17116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.30115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.45109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
120deepseek/deepseek-v4-prohigh108.2031
3 columns on 1 board, hardest first are the whole evidence behind 108.20. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.16119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
121google/gemini-3-flash-previewunstated108.1691
9 columns on 1 board, hardest first are the whole evidence behind 108.16. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.17116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.36115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.51109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7584 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
123doubao-seed-2.0-pro-260215not in the cataloghigh108.1241
4 columns on 1 board, hardest first are the whole evidence behind 108.12. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.20119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.39115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
122gpt-5-2025-08-07not in the catalogmedium108.12101
10 columns on 1 board, hardest first are the whole evidence behind 108.12. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.15121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.06120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.32119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.27115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.70106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
124thinkingmachines/inklingxhigh107.98292
29 columns on 2 boards, hardest first are the whole evidence behind 107.98. Each is a door to the column and everything else scored on it.
- typescriptLiveBench30122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- tablejoinLiveBench48.48117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.05116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.40115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.33109 on the scale
- plot_unscramblingLiveBench53.21106 on the scale
- integrals_with_gameLiveBench72104 on the scale
- consecutive_eventsLiveBench69.85102 on the scale
- zebra_puzzleLiveBench88102 on the scale
- javascriptLiveBench68.1899 on the scale
- logic_with_navigationLiveBench6696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- simplifyLiveBench60.5092 on the scale
- connectionsLiveBench91.1791 on the scale
- theory_of_mindLiveBench65.3988 on the scale
- code_completionLiveBench67.3987 on the scale
- paraphraseLiveBench70.8586 on the scale
- summarizeLiveBench74.8585 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
- typosLiveBench7677 on the scale
- code_generationLiveBench74.6576 on the scale
- story_generationLiveBench74.1874 on the scale
- olympiadLiveBench85.4072 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9947 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench10041 on the scale
125openai/gpt-5.5low107.8641
4 columns on 1 board, hardest first are the whole evidence behind 107.86. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.28121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.30119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
127google/gemini-3.6-flashlow107.8041
4 columns on 1 board, hardest first are the whole evidence behind 107.80. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.23121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.42119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
126qwen/qwen3.7-plusunstated107.8031
3 columns on 1 board, hardest first are the whole evidence behind 107.80. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
128moonshotai/kimi-k2.7-codeunstated107.73292
29 columns on 2 boards, hardest first are the whole evidence behind 107.73. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.14119 on the scale
- tablejoinLiveBench47.54117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.12116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.36115 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.54109 on the scale
- plot_unscramblingLiveBench55.72106 on the scale
- integrals_with_gameLiveBench37104 on the scale
- consecutive_eventsLiveBench48.28102 on the scale
- zebra_puzzleLiveBench96102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench7496 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9694 on the scale
- simplifyLiveBench49.7592 on the scale
- connectionsLiveBench9891 on the scale
- theory_of_mindLiveBench69.2388 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench48.5386 on the scale
- summarizeLiveBench58.3285 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
- typosLiveBench8077 on the scale
- code_generationLiveBench71.8376 on the scale
- story_generationLiveBench68.5774 on the scale
- olympiadLiveBench90.3072 on the scale
- spatialLiveBench9253 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench92.1641 on the scale
129x-ai/grok-4.3high107.3961
6 columns on 1 board, hardest first are the whole evidence behind 107.39. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.18119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.15116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.33115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.43109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
130fireworks/kimi-k2p5not in the catalogunstated107.3241
4 columns on 1 board, hardest first are the whole evidence behind 107.32. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.28115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
131kimi-k2.6-thinkingnot in the catalogunstated107.26231
23 columns on 1 board, hardest first are the whole evidence behind 107.26. Each is a door to the column and everything else scored on it.
- typescriptLiveBench26.67122 on the scale
- tablejoinLiveBench46.10117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench58.09106 on the scale
- integrals_with_gameLiveBench54104 on the scale
- consecutive_eventsLiveBench51.27102 on the scale
- zebra_puzzleLiveBench78.50102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench61.4292 on the scale
- connectionsLiveBench89.3391 on the scale
- theory_of_mindLiveBench7588 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench61.6386 on the scale
- summarizeLiveBench67.5085 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench78.8776 on the scale
- story_generationLiveBench66.8874 on the scale
- olympiadLiveBench90.0372 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench98.0441 on the scale
132deepseek/deepseek-v4-promax107.0271
7 columns on 1 board, hardest first are the whole evidence behind 107.02. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.19119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.02116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.47115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.45109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9794 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7884 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9083 on the scale
133openai/gpt-5.6-terralow106.9341
4 columns on 1 board, hardest first are the whole evidence behind 106.93. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.19121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8783 on the scale
134gpt-5.4-mini-2026-03-17not in the catalogxhigh106.9151
5 columns on 1 board, hardest first are the whole evidence behind 106.91. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.17119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.10116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.51109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8783 on the scale
135gpt-5.4-nano-2026-03-17not in the cataloghigh106.76101
10 columns on 1 board, hardest first are the whole evidence behind 106.76. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.05121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.06120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.12116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.12115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.26115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.45109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.60106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7883 on the scale
136qwen3.6-max-previewnot in the catalogunstated106.7581
8 columns on 1 board, hardest first are the whole evidence behind 106.75. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.52115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.23115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.50106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7784 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8783 on the scale
137thinkingmachines/inkling-smallxhigh106.5461
6 columns on 1 board, hardest first are the whole evidence behind 106.54. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.06121 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.17116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.19115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.46109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8983 on the scale
138qwen/qwen3.6-plusunstated106.49332
33 columns on 2 boards, hardest first are the whole evidence behind 106.49. Each is a door to the column and everything else scored on it.
- typescriptLiveBench20122 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.08120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.16119 on the scale
- tablejoinLiveBench44.65117 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.44115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.26115 on the scale
- pythonLiveBench45109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.38109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.50106 on the scale
- plot_unscramblingLiveBench52.47106 on the scale
- integrals_with_gameLiveBench62104 on the scale
- consecutive_eventsLiveBench67.04102 on the scale
- zebra_puzzleLiveBench68102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7096 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9394 on the scale
- simplifyLiveBench51.5792 on the scale
- connectionsLiveBench90.5091 on the scale
- theory_of_mindLiveBench67.3188 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench60.6286 on the scale
- summarizeLiveBench64.3285 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.5884 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
- typosLiveBench8277 on the scale
- code_generationLiveBench80.2876 on the scale
- story_generationLiveBench56.8774 on the scale
- olympiadLiveBench82.7672 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench93.1442 on the scale
- tablereformatLiveBench98.0441 on the scale
139doubao-seed-2.0-lite-260428not in the cataloghigh106.4241
4 columns on 1 board, hardest first are the whole evidence behind 106.42. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.20119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.32115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
140x-ai/grok-build-0.1unstated106.25231
23 columns on 1 board, hardest first are the whole evidence behind 106.25. Each is a door to the column and everything else scored on it.
- typescriptLiveBench33.33122 on the scale
- tablejoinLiveBench42.65117 on the scale
- pythonLiveBench45109 on the scale
- plot_unscramblingLiveBench53.38106 on the scale
- integrals_with_gameLiveBench66104 on the scale
- consecutive_eventsLiveBench69.73102 on the scale
- zebra_puzzleLiveBench68.25102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench7096 on the scale
- simplifyLiveBench62.6092 on the scale
- connectionsLiveBench8491 on the scale
- theory_of_mindLiveBench69.2388 on the scale
- code_completionLiveBench67.3987 on the scale
- paraphraseLiveBench58.3886 on the scale
- summarizeLiveBench67.3785 on the scale
- typosLiveBench8077 on the scale
- code_generationLiveBench63.3876 on the scale
- story_generationLiveBench72.5374 on the scale
- olympiadLiveBench86.6272 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench6647 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench10041 on the scale
141gpt-5.4-mini-2026-03-17not in the cataloghigh106.0971
7 columns on 1 board, hardest first are the whole evidence behind 106.09. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.23119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.29115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.28115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.50106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8483 on the scale
142ox-alphanot in the catalogmax106.05231
23 columns on 1 board, hardest first are the whole evidence behind 106.05. Each is a door to the column and everything else scored on it.
- typescriptLiveBench43.33122 on the scale
- tablejoinLiveBench46.94117 on the scale
- pythonLiveBench60109 on the scale
- plot_unscramblingLiveBench53.08106 on the scale
- integrals_with_gameLiveBench39104 on the scale
- consecutive_eventsLiveBench80.35102 on the scale
- zebra_puzzleLiveBench79.50102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench5296 on the scale
- simplifyLiveBench54.4892 on the scale
- connectionsLiveBench71.3391 on the scale
- theory_of_mindLiveBench78.8588 on the scale
- code_completionLiveBench78.2687 on the scale
- paraphraseLiveBench57.4786 on the scale
- summarizeLiveBench58.1885 on the scale
- typosLiveBench7477 on the scale
- code_generationLiveBench73.2476 on the scale
- story_generationLiveBench70.8774 on the scale
- olympiadLiveBench81.9572 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench91.1842 on the scale
- tablereformatLiveBench10041 on the scale
143nvidia/nemotron-3-ultra-550b-a55bunstated105.72231
23 columns on 1 board, hardest first are the whole evidence behind 105.72. Each is a door to the column and everything else scored on it.
- typescriptLiveBench16.67122 on the scale
- tablejoinLiveBench46.56117 on the scale
- pythonLiveBench45109 on the scale
- plot_unscramblingLiveBench47.77106 on the scale
- integrals_with_gameLiveBench80104 on the scale
- consecutive_eventsLiveBench18.77102 on the scale
- zebra_puzzleLiveBench71.25102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench6896 on the scale
- simplifyLiveBench72.2892 on the scale
- connectionsLiveBench86.6791 on the scale
- theory_of_mindLiveBench61.5488 on the scale
- code_completionLiveBench69.5687 on the scale
- paraphraseLiveBench77.3886 on the scale
- summarizeLiveBench74.8885 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench71.8376 on the scale
- story_generationLiveBench69.2374 on the scale
- olympiadLiveBench82.5872 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9647 on the scale
- math_compLiveBench96.0842 on the scale
- tablereformatLiveBench98.0441 on the scale
144nemotron-3-ultranot in the catalogunstated105.6231
3 columns on 1 board, hardest first are the whole evidence behind 105.62. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.20121 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
145google/gemini-3.6-flashminimal105.5541
4 columns on 1 board, hardest first are the whole evidence behind 105.55. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.25121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.29119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
146z-ai/glm-5.2low105.4251
5 columns on 1 board, hardest first are the whole evidence behind 105.42. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.19121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.18119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.55109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
147gpt-5.1-2025-11-13not in the catalogmedium105.4151
5 columns on 1 board, hardest first are the whole evidence behind 105.41. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.16121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.27115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
148qwen3.5-plusnot in the catalogunstated105.3181
8 columns on 1 board, hardest first are the whole evidence behind 105.31. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.16121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.09119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.25115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.21115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.50106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
149gpt-5-2025-08-07not in the cataloglow105.2231
3 columns on 1 board, hardest first are the whole evidence behind 105.22. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.16121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.37109 on the scale
150gpt-5.2-2025-12-11not in the cataloglow104.8991
9 columns on 1 board, hardest first are the whole evidence behind 104.89. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.10121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.06120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.26119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.33115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.27115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
151moonshotai/kimi-k2-thinkingnot in the catalogunstated104.8241
4 columns on 1 board, hardest first are the whole evidence behind 104.82. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.21115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
152claude-opus-4-5-20251101-16knot in the catalogunstated104.7051
5 columns on 1 board, hardest first are the whole evidence behind 104.70. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.20115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
153minimax/minimax-m3unstated104.58272
27 columns on 2 boards, hardest first are the whole evidence behind 104.58. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.07121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.17119 on the scale
- tablejoinLiveBench45.85117 on the scale
- pythonLiveBench35109 on the scale
- plot_unscramblingLiveBench50.17106 on the scale
- integrals_with_gameLiveBench56104 on the scale
- consecutive_eventsLiveBench84.61102 on the scale
- zebra_puzzleLiveBench61102 on the scale
- javascriptLiveBench63.6499 on the scale
- logic_with_navigationLiveBench6696 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7194 on the scale
- simplifyLiveBench56.6292 on the scale
- connectionsLiveBench96.3391 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench67.3987 on the scale
- paraphraseLiveBench55.9386 on the scale
- summarizeLiveBench53.9385 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.9183 on the scale
- typosLiveBench8477 on the scale
- code_generationLiveBench69.0176 on the scale
- story_generationLiveBench63.5574 on the scale
- olympiadLiveBench85.6172 on the scale
- spatialLiveBench9453 on the scale
- AMPS_HardLiveBench7547 on the scale
- math_compLiveBench91.1842 on the scale
- tablereformatLiveBench98.0441 on the scale
154gpt-5-mini-2025-08-07not in the cataloghigh104.57121
12 columns on 1 board, hardest first are the whole evidence behind 104.57. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.05121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.06120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.24119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.12116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.22115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.27115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.47109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.30106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
155google/gemini-3.5-flashminimal104.4131
3 columns on 1 board, hardest first are the whole evidence behind 104.41. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.27119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
156qwen/qwen3.5-397b-a17bunstated104.4031
3 columns on 1 board, hardest first are the whole evidence behind 104.40. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.29109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
157claude-opus-4-8not in the catalognone104.3131
3 columns on 1 board, hardest first are the whole evidence behind 104.31. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.19119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
158openai/gpt-5.4-minixhigh104.22231
23 columns on 1 board, hardest first are the whole evidence behind 104.22. Each is a door to the column and everything else scored on it.
- typescriptLiveBench20122 on the scale
- tablejoinLiveBench49.38117 on the scale
- pythonLiveBench55109 on the scale
- plot_unscramblingLiveBench43.88106 on the scale
- integrals_with_gameLiveBench37104 on the scale
- consecutive_eventsLiveBench62.97102 on the scale
- zebra_puzzleLiveBench60.37102 on the scale
- javascriptLiveBench5099 on the scale
- logic_with_navigationLiveBench6096 on the scale
- simplifyLiveBench54.0792 on the scale
- connectionsLiveBench98.9891 on the scale
- theory_of_mindLiveBench76.9288 on the scale
- code_completionLiveBench75.5687 on the scale
- paraphraseLiveBench58.8886 on the scale
- summarizeLiveBench59.6385 on the scale
- typosLiveBench7077 on the scale
- code_generationLiveBench67.6976 on the scale
- story_generationLiveBench66.6374 on the scale
- olympiadLiveBench85.7372 on the scale
- spatialLiveBench8853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench94.1242 on the scale
- tablereformatLiveBench10041 on the scale
159gpt-5.4-2026-03-05not in the cataloglow104.1641
4 columns on 1 board, hardest first are the whole evidence behind 104.16. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.19119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
160grok-4-0709not in the catalogunstated104.1651
5 columns on 1 board, hardest first are the whole evidence behind 104.16. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.20115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8783 on the scale
161qwen/qwen3-235b-a22b-thinking-2507unstated104.1041
4 columns on 1 board, hardest first are the whole evidence behind 104.10. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.09121 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.40115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8083 on the scale
162z-ai/glm-4.7unstated104.0741
4 columns on 1 board, hardest first are the whole evidence behind 104.07. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.32115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
163claude-opus-4-5-20251101-32knot in the catalogunstated104.0581
8 columns on 1 board, hardest first are the whole evidence behind 104.05. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.05119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.05116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.46115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.21115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.34109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
164gpt-5-mini-2025-08-07not in the catalogmedium103.8391
9 columns on 1 board, hardest first are the whole evidence behind 103.83. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.04121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.20115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7894 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.6584 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9781 on the scale
165google/gemini-2.5-prounstated103.7391
9 columns on 1 board, hardest first are the whole evidence behind 103.73. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.14115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.25109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.5884 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
166qwen/qwen3.6-27bunstated103.72262
26 columns on 2 boards, hardest first are the whole evidence behind 103.72. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- tablejoinLiveBench42.85117 on the scale
- pythonLiveBench40109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.35109 on the scale
- plot_unscramblingLiveBench42.75106 on the scale
- integrals_with_gameLiveBench52104 on the scale
- consecutive_eventsLiveBench70.39102 on the scale
- zebra_puzzleLiveBench53.75102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench6296 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.9194 on the scale
- simplifyLiveBench50.7392 on the scale
- connectionsLiveBench77.1791 on the scale
- theory_of_mindLiveBench65.3988 on the scale
- code_completionLiveBench71.7487 on the scale
- paraphraseLiveBench49.4386 on the scale
- summarizeLiveBench52.6285 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
- typosLiveBench7077 on the scale
- code_generationLiveBench71.8376 on the scale
- story_generationLiveBench60.1374 on the scale
- olympiadLiveBench82.3272 on the scale
- spatialLiveBench10053 on the scale
- AMPS_HardLiveBench9347 on the scale
- math_compLiveBench92.1642 on the scale
- tablereformatLiveBench98.0441 on the scale
167qwen/qwen3.5-397b-a17bnone103.5541
4 columns on 1 board, hardest first are the whole evidence behind 103.55. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.18121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.31109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8683 on the scale
168o4-mini-2025-04-16not in the cataloghigh103.35121
12 columns on 1 board, hardest first are the whole evidence behind 103.35. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.05121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.06120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.13119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.05116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.19115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.25115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.36109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.30106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
169gemini-2.5-pro-preview-06-05not in the catalogunstated103.2751
5 columns on 1 board, hardest first are the whole evidence behind 103.27. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.10115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.30106 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
170qwen/qwen3.7-plusnone103.1741
4 columns on 1 board, hardest first are the whole evidence behind 103.17. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.34109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
171x-ai/grok-4.3unstated103.07231
23 columns on 1 board, hardest first are the whole evidence behind 103.07. Each is a door to the column and everything else scored on it.
- typescriptLiveBench13.33122 on the scale
- tablejoinLiveBench41.25117 on the scale
- pythonLiveBench15109 on the scale
- plot_unscramblingLiveBench48.74106 on the scale
- integrals_with_gameLiveBench63104 on the scale
- consecutive_eventsLiveBench28.03102 on the scale
- zebra_puzzleLiveBench57.75102 on the scale
- javascriptLiveBench27.2799 on the scale
- logic_with_navigationLiveBench6696 on the scale
- simplifyLiveBench57.6092 on the scale
- connectionsLiveBench9491 on the scale
- theory_of_mindLiveBench61.5488 on the scale
- code_completionLiveBench65.2287 on the scale
- paraphraseLiveBench64.3886 on the scale
- summarizeLiveBench58.4285 on the scale
- typosLiveBench7877 on the scale
- code_generationLiveBench74.6576 on the scale
- story_generationLiveBench70.6074 on the scale
- olympiadLiveBench82.2672 on the scale
- spatialLiveBench9853 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench95.1042 on the scale
- tablereformatLiveBench98.0441 on the scale
172claude-sonnet-4-6not in the catalogmedium103.0131
3 columns on 1 board, hardest first are the whole evidence behind 103.01. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.16119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
174deepseek/deepseek-v4-flashunstated102.96231
23 columns on 1 board, hardest first are the whole evidence behind 102.96. Each is a door to the column and everything else scored on it.
- typescriptLiveBench23.33122 on the scale
- tablejoinLiveBench46.12117 on the scale
- pythonLiveBench35109 on the scale
- plot_unscramblingLiveBench46.87106 on the scale
- integrals_with_gameLiveBench37104 on the scale
- consecutive_eventsLiveBench59.91102 on the scale
- zebra_puzzleLiveBench49.25102 on the scale
- javascriptLiveBench54.5599 on the scale
- logic_with_navigationLiveBench6496 on the scale
- simplifyLiveBench54.0592 on the scale
- connectionsLiveBench89.5091 on the scale
- theory_of_mindLiveBench73.0888 on the scale
- code_completionLiveBench65.2287 on the scale
- paraphraseLiveBench62.3886 on the scale
- summarizeLiveBench68.8585 on the scale
- typosLiveBench7477 on the scale
- code_generationLiveBench73.2476 on the scale
- story_generationLiveBench67.2774 on the scale
- olympiadLiveBench86.5372 on the scale
- spatialLiveBench9653 on the scale
- AMPS_HardLiveBench9847 on the scale
- math_compLiveBench97.0642 on the scale
- tablereformatLiveBench98.0441 on the scale
173o3-2025-04-16not in the cataloghigh102.96111
11 columns on 1 board, hardest first are the whole evidence behind 102.96. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.29121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.49115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.19115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.33109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.10106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
175kimi-k2-thinking-turbonot in the catalogunstated102.4631
3 columns on 1 board, hardest first are the whole evidence behind 102.46. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8483 on the scale
176o3-mini-2025-01-31not in the cataloghigh102.45121
12 columns on 1 board, hardest first are the whole evidence behind 102.45. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.07121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.15119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.15115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.12115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.19109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.40106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9681 on the scale
177gemini-2.5-pro-preview-05-06not in the catalogunstated102.4332
3 columns on 2 boards, hardest first are the whole evidence behind 102.43. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot76.9097 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9681 on the scale
178claude-sonnet-4-6not in the cataloghigh102.3041
4 columns on 1 board, hardest first are the whole evidence behind 102.30. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.19119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.35115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
179qwen/qwen3.7-flashunstated102.1641
4 columns on 1 board, hardest first are the whole evidence behind 102.16. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.15121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.19109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
180google/gemini-3.1-flash-litehigh102.1331
3 columns on 1 board, hardest first are the whole evidence behind 102.13. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.28109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
181qwen/qwen3.6-35b-a3bunstated102.0431
3 columns on 1 board, hardest first are the whole evidence behind 102.04. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.18109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8483 on the scale
182o4-mini-2025-04-16not in the catalogmedium101.9381
8 columns on 1 board, hardest first are the whole evidence behind 101.93. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.22119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.19115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.29109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7883 on the scale
183openai/gpt-5.6-lunalow101.9251
5 columns on 1 board, hardest first are the whole evidence behind 101.92. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.17119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.41109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
184qwen/qwen3-235b-a22b-thinking-2507not in the catalogunstated101.2941
4 columns on 1 board, hardest first are the whole evidence behind 101.29. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.08115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
185claude-sonnet-4-5-20250929-59knot in the catalogunstated101.2651
5 columns on 1 board, hardest first are the whole evidence behind 101.26. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.10119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.31115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.13115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
186claude-sonnet-4-5-20250929-32knot in the catalogunstated101.1881
8 columns on 1 board, hardest first are the whole evidence behind 101.18. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.10119 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.02116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.15115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.24109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9881 on the scale
187claude-sonnet-4-6not in the catalogmax101.1741
4 columns on 1 board, hardest first are the whole evidence behind 101.17. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.23119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.33115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7983 on the scale
188o3-2025-04-16not in the catalogmedium101.1181
8 columns on 1 board, hardest first are the whole evidence behind 101.11. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.23121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.37119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.17115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.30109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.6284 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8183 on the scale
189openai/gpt-5.6-solnone100.9841
4 columns on 1 board, hardest first are the whole evidence behind 100.98. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.33121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.07119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
190qwen/qwen3.6-flashunstated100.9591
9 columns on 1 board, hardest first are the whole evidence behind 100.95. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.12121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.12119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.16115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.10115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.17109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
191gpt-5.1-2025-11-13not in the cataloglow100.6441
4 columns on 1 board, hardest first are the whole evidence behind 100.64. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.19121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.23119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.17115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6494 on the scale
192o4-mini-2025-04-16not in the cataloglow100.6171
7 columns on 1 board, hardest first are the whole evidence behind 100.61. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.15119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.20115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.11115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.16109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.30106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7583 on the scale
193qwen/qwen3.7-flashnone100.5441
4 columns on 1 board, hardest first are the whole evidence behind 100.54. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.14121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.19109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8183 on the scale
194gpt-5-nano-2025-08-07not in the cataloghigh100.45121
12 columns on 1 board, hardest first are the whole evidence behind 100.45. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.08121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.22119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.02116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.12115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.08115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.20109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9581 on the scale
195qwen/qwen3.6-27bnone100.3241
4 columns on 1 board, hardest first are the whole evidence behind 100.32. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.07121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.34109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
196qwen/qwen3.6-35b-a3bnone100.2241
4 columns on 1 board, hardest first are the whole evidence behind 100.22. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.22121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.20109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
197moonshotai/kimi-k3low100.1241
4 columns on 1 board, hardest first are the whole evidence behind 100.12. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.12121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.18119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8583 on the scale
198qwen3-max-2025-09-23not in the catalogunstated100.0361
6 columns on 1 board, hardest first are the whole evidence behind 100.03. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.05121 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.49115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.19109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7383 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9781 on the scale
199google/gemini-3.5-flash-litehigh99.97282
28 columns on 2 boards, hardest first are the whole evidence behind 99.97. Each is a door to the column and everything else scored on it.
- typescriptLiveBench26.67122 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.16119 on the scale
- tablejoinLiveBench40.62117 on the scale
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0116 on the scale
- pythonLiveBench50109 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.26109 on the scale
- plot_unscramblingLiveBench49.46106 on the scale
- integrals_with_gameLiveBench26104 on the scale
- consecutive_eventsLiveBench21.09102 on the scale
- zebra_puzzleLiveBench32.75102 on the scale
- javascriptLiveBench59.0999 on the scale
- logic_with_navigationLiveBench6896 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7194 on the scale
- simplifyLiveBench68.0292 on the scale
- connectionsLiveBench9291 on the scale
- theory_of_mindLiveBench5088 on the scale
- code_completionLiveBench76.0987 on the scale
- paraphraseLiveBench6586 on the scale
- summarizeLiveBench68.4785 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
- typosLiveBench7477 on the scale
- code_generationLiveBench76.0676 on the scale
- story_generationLiveBench67.4774 on the scale
- olympiadLiveBench81.8072 on the scale
- spatialLiveBench9053 on the scale
- AMPS_HardLiveBench9547 on the scale
- math_compLiveBench92.1642 on the scale
- tablereformatLiveBench98.0441 on the scale
200z-ai/glm-5unstated99.9671
7 columns on 1 board, hardest first are the whole evidence behind 99.96. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.16115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.02106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8094 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7284 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8883 on the scale
201claude-opus-4-5-20251101not in the catalogunstated99.9271
7 columns on 1 board, hardest first are the whole evidence behind 99.92. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.05119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.21115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.20106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4894 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7784 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8183 on the scale
202gpt-5.5-instantnot in the catalogunstated99.8941
4 columns on 1 board, hardest first are the whole evidence behind 99.89. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub0.02116 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.26109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8383 on the scale
203qwen3.5-flashnot in the catalogunstated99.5891
9 columns on 1 board, hardest first are the whole evidence behind 99.58. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.20121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.14119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.20115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.09109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.10106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.8494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8283 on the scale
204o3-2025-04-16not in the cataloglow99.3761
6 columns on 1 board, hardest first are the whole evidence behind 99.37. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.36119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.10115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.19109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.10106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8083 on the scale
205claude-sonnet-4-20250514-59knot in the catalogunstated99.1731
3 columns on 1 board, hardest first are the whole evidence behind 99.17. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7883 on the scale
206gpt-5-nano-2025-08-07not in the catalogmedium99.0381
8 columns on 1 board, hardest first are the whole evidence behind 99.03. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.09121 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.07115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.10106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9581 on the scale
207deepseek/deepseek-r1-0528unstated98.8141
4 columns on 1 board, hardest first are the whole evidence behind 98.81. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9781 on the scale
208google/gemini-3.5-flash-litelow98.6841
4 columns on 1 board, hardest first are the whole evidence behind 98.68. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.19121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.21119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7683 on the scale
209gpt-5.4-nano-2026-03-17not in the cataloglow98.4051
5 columns on 1 board, hardest first are the whole evidence behind 98.40. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.03121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.20119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.20109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7283 on the scale
210o1-2024-12-17not in the cataloghigh98.2792
9 columns on 2 boards, hardest first are the whole evidence behind 98.27. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.10119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.41115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.09115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.15109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot61.7097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9581 on the scale
211o3-mini-2025-01-31not in the catalogmedium97.9471
7 columns on 1 board, hardest first are the whole evidence behind 97.94. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.11119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.08115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.11109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.11106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9581 on the scale
212o1-2024-12-17not in the catalogmedium97.8151
5 columns on 1 board, hardest first are the whole evidence behind 97.81. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.08119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.10109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9481 on the scale
213seed-1.6not in the catalogunstated97.7041
4 columns on 1 board, hardest first are the whole evidence behind 97.70. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.05119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.17115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7883 on the scale
214claude-opus-4-20250514-27knot in the catalogunstated97.6941
4 columns on 1 board, hardest first are the whole evidence behind 97.69. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6494 on the scale
215claude-opus-4-1-20250805-27knot in the catalogunstated97.2951
5 columns on 1 board, hardest first are the whole evidence behind 97.29. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.04120 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.07115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
216openai/gpt-5.5none97.1141
4 columns on 1 board, hardest first are the whole evidence behind 97.11. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.18121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.09119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
217grok-3-mini-betanot in the cataloghigh97.0372
7 columns on 2 boards, hardest first are the whole evidence behind 97.03. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot49.3097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.7894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8881 on the scale
218gpt-5.2-2025-12-11not in the catalognone96.5341
4 columns on 1 board, hardest first are the whole evidence behind 96.53. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.14121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.02119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7383 on the scale
219claude-opus-4-1-20250805-16knot in the catalogunstated96.3931
3 columns on 1 board, hardest first are the whole evidence behind 96.39. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
220z-ai/glm-5.2none96.2251
5 columns on 1 board, hardest first are the whole evidence behind 96.22. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.18121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.19119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.42109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7183 on the scale
221qwen/qwen3-32bunstated96.2132
3 columns on 2 boards, hardest first are the whole evidence behind 96.21. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot4097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6683 on the scale
222gpt-5.4-2026-03-05not in the catalognone96.1541
4 columns on 1 board, hardest first are the whole evidence behind 96.15. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.16121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.03119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7583 on the scale
223deepseek-r1not in the catalogunstated96.0842
4 columns on 2 boards, hardest first are the whole evidence behind 96.08. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot56.9097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9381 on the scale
224claude-haiku-4-5-20251001-32knot in the catalogunstated96.0181
8 columns on 1 board, hardest first are the whole evidence behind 96.01. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.05119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.13115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9681 on the scale
225openai/gpt-5.6-lunanone95.9351
5 columns on 1 board, hardest first are the whole evidence behind 95.93. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.20121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.03119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.40109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6483 on the scale
226openai/gpt-5.6-terranone95.5741
4 columns on 1 board, hardest first are the whole evidence behind 95.57. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.14121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.05119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
227claude-3-7-sonnet-20250219-64knot in the catalogunstated95.1861
6 columns on 1 board, hardest first are the whole evidence behind 95.18. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.03115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9181 on the scale
228gpt-5-2025-08-07not in the catalogminimal95.1451
5 columns on 1 board, hardest first are the whole evidence behind 95.14. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.14121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.11119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.18109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7283 on the scale
229google/gemini-3.5-flash-liteminimal94.9541
4 columns on 1 board, hardest first are the whole evidence behind 94.95. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.12121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.07119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
230google/gemini-3.1-flash-litelow94.9431
3 columns on 1 board, hardest first are the whole evidence behind 94.94. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.22109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
232grok-3-mini-betanot in the cataloglow94.7162
6 columns on 2 boards, hardest first are the whole evidence behind 94.71. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.03115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot34.7097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.6294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9181 on the scale
231o1-2024-12-17not in the cataloglow94.7141
4 columns on 1 board, hardest first are the whole evidence behind 94.71. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.11119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.08109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
233grok-3-betanot in the catalogunstated94.6582
8 columns on 2 boards, hardest first are the whole evidence behind 94.65. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot53.3097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8981 on the scale
234claude-3-7-sonnet-20250219-32knot in the catalogunstated94.4051
5 columns on 1 board, hardest first are the whole evidence behind 94.40. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.03115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9081 on the scale
235deepseek/deepseek-v4-pronone94.2841
4 columns on 1 board, hardest first are the whole evidence behind 94.28. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.17121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.04119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7383 on the scale
236gpt-5-mini-2025-08-07not in the catalogminimal93.8651
5 columns on 1 board, hardest first are the whole evidence behind 93.86. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.10121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.06109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7283 on the scale
237claude-opus-4-1-20250805not in the catalogunstated93.8161
6 columns on 1 board, hardest first are the whole evidence behind 93.81. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.08123 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4094 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7384 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7383 on the scale
238google/gemini-3.1-flash-liteminimal93.6831
3 columns on 1 board, hardest first are the whole evidence behind 93.68. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.21109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
239deepseek-r1-distill-llama-70bnot in the catalogunstated93.5531
3 columns on 1 board, hardest first are the whole evidence behind 93.55. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.9081 on the scale
240gpt-4.1-mini-2025-04-14not in the catalogunstated93.4891
9 columns on 1 board, hardest first are the whole evidence behind 93.48. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.07121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.01119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.13115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.07109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0.10106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8781 on the scale
241claude-sonnet-4-5-20250929not in the catalogunstated93.21101
10 columns on 1 board, hardest first are the whole evidence behind 93.21. Each is a door to the column and everything else scored on it.
- EBR-benchEpoch AI Benchmarking Hub0.02123 on the scale
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.06119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.24115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.09115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3694 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7483 on the scale
242claude-3-7-sonnet-20250219-16knot in the catalogunstated93.2051
5 columns on 1 board, hardest first are the whole evidence behind 93.20. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8681 on the scale
243qwq-32bnot in the catalogunstated93.1832
3 columns on 2 boards, hardest first are the whole evidence behind 93.18. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot20.9097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6583 on the scale
244gpt-5-nano-2025-08-07not in the cataloglow92.7451
5 columns on 1 board, hardest first are the whole evidence behind 92.74. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.09121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.15119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.06109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5883 on the scale
245o1-mini-2024-09-12not in the cataloghigh92.5351
5 columns on 1 board, hardest first are the whole evidence behind 92.53. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8981 on the scale
246claude-opus-4-20250514not in the catalogunstated92.4881
8 columns on 1 board, hardest first are the whole evidence behind 92.48. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4294 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.7184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8581 on the scale
247deepseek-r1-distill-qwen-14bnot in the catalogunstated92.3031
3 columns on 1 board, hardest first are the whole evidence behind 92.30. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8781 on the scale
248zai-org/glm-4.6not in the catalogunstated92.2141
4 columns on 1 board, hardest first are the whole evidence behind 92.21. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0.02120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
249gemini-2.0-flash-thinking-exp-01-21not in the catalogunstated92.1432
3 columns on 2 boards, hardest first are the whole evidence behind 92.14. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot18.2097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.5894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5783 on the scale
250o3-mini-2025-01-31not in the cataloglow91.9141
4 columns on 1 board, hardest first are the whole evidence behind 91.91. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.06119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.04109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6883 on the scale
251gpt-5.1-2025-11-13not in the catalognone91.9051
5 columns on 1 board, hardest first are the whole evidence behind 91.90. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.15121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.06119 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.02115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
252gpt-4.1-2025-04-14not in the catalogunstated91.64111
11 columns on 1 board, hardest first are the whole evidence behind 91.64. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.11119 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.31115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.06109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3894 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.4984 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8381 on the scale
253o1-mini-2024-09-12not in the catalogmedium91.5151
5 columns on 1 board, hardest first are the whole evidence behind 91.51. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.02115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8481 on the scale
254gpt-5.4-nano-2026-03-17not in the catalognone91.3351
5 columns on 1 board, hardest first are the whole evidence behind 91.33. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.09121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.02119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.05109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.4794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5683 on the scale
255claude-haiku-4-5-20251001not in the catalogunstated90.9951
5 columns on 1 board, hardest first are the whole evidence behind 90.99. Each is a door to the column and everything else scored on it.
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.13115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8781 on the scale
257minimax/minimax-m3none90.8441
4 columns on 1 board, hardest first are the whole evidence behind 90.84. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.08121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.08119 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.8183 on the scale
256gemini-2.0-pro-exp-02-05not in the catalogunstated90.8442
4 columns on 2 boards, hardest first are the whole evidence behind 90.84. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot35.6097 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8381 on the scale
258gpt-4.5-preview-2025-02-27not in the catalogunstated90.7531
3 columns on 1 board, hardest first are the whole evidence behind 90.75. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7981 on the scale
259gpt-5.4-mini-2026-03-17not in the catalognone90.4851
5 columns on 1 board, hardest first are the whole evidence behind 90.48. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.11121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.02119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.17109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6483 on the scale
260claude-sonnet-4-20250514not in the catalogunstated90.1671
7 columns on 1 board, hardest first are the whole evidence behind 90.16. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.04115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8481 on the scale
261deepseek-v3-0324not in the catalogunstated90.1241
4 columns on 1 board, hardest first are the whole evidence behind 90.12. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7681 on the scale
262gemini-2.0-flash-001not in the catalogunstated89.7551
5 columns on 1 board, hardest first are the whole evidence behind 89.75. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.02115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8281 on the scale
263mistral-medium-2505not in the catalogunstated89.4151
5 columns on 1 board, hardest first are the whole evidence behind 89.41. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.00115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8281 on the scale
264o1-preview-2024-09-12not in the catalogunstated89.0831
3 columns on 1 board, hardest first are the whole evidence behind 89.08. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.8281 on the scale
265magistral-small-2506not in the catalogunstated88.6331
3 columns on 1 board, hardest first are the whole evidence behind 88.63. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5683 on the scale
266gpt-5-nano-2025-08-07not in the catalogminimal88.0851
5 columns on 1 board, hardest first are the whole evidence behind 88.08. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.05121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.02119 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.02109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.3694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4883 on the scale
267qwen3-235b-a22bnot in the catalogunstated87.5631
3 columns on 1 board, hardest first are the whole evidence behind 87.56. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.7183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6981 on the scale
268nemotron-3.5-lightning-30b-a3bnot in the catalogunstated87.31231
23 columns on 1 board, hardest first are the whole evidence behind 87.31. Each is a door to the column and everything else scored on it.
- typescriptLiveBench6.67122 on the scale
- tablejoinLiveBench40.48117 on the scale
- pythonLiveBench35109 on the scale
- plot_unscramblingLiveBench22.87106 on the scale
- integrals_with_gameLiveBench24104 on the scale
- consecutive_eventsLiveBench2.51102 on the scale
- zebra_puzzleLiveBench17.25102 on the scale
- javascriptLiveBench18.1899 on the scale
- logic_with_navigationLiveBench1896 on the scale
- simplifyLiveBench52.6892 on the scale
- connectionsLiveBench3491 on the scale
- theory_of_mindLiveBench44.2388 on the scale
- code_completionLiveBench23.9187 on the scale
- paraphraseLiveBench57.3286 on the scale
- summarizeLiveBench50.4285 on the scale
- typosLiveBench5877 on the scale
- code_generationLiveBench67.6176 on the scale
- story_generationLiveBench59.3574 on the scale
- olympiadLiveBench58.9272 on the scale
- spatialLiveBench9253 on the scale
- AMPS_HardLiveBench9747 on the scale
- math_compLiveBench86.2842 on the scale
- tablereformatLiveBench86.2841 on the scale
269claude-3-7-sonnet-20250219not in the catalogunstated87.2261
6 columns on 1 board, hardest first are the whole evidence behind 87.22. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.03115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2294 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.6184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6881 on the scale
270llama-4-maverick-17b-128e-instruct-fp8not in the catalogunstated87.2151
5 columns on 1 board, hardest first are the whole evidence behind 87.21. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.6783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7381 on the scale
271gemini-1.5-pro-002not in the catalogunstated86.8031
3 columns on 1 board, hardest first are the whole evidence behind 86.80. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7081 on the scale
272gpt-4.1-nano-2025-04-14not in the catalogunstated86.2271
7 columns on 1 board, hardest first are the whole evidence behind 86.22. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.06115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2994 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7081 on the scale
273claude-3-5-sonnet-20241022not in the catalogunstated85.3182
8 columns on 2 boards, hardest first are the whole evidence behind 85.31. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.02115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot51.6097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5781 on the scale
274qwen-max-2025-01-25not in the catalogunstated85.1162
6 columns on 2 boards, hardest first are the whole evidence behind 85.11. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot21.8097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6781 on the scale
275google/gemma-3-27b-itunstated85.0542
4 columns on 2 boards, hardest first are the whole evidence behind 85.05. Each is a door to the column and everything else scored on it.
- polyglot pass rateAider polyglot4.9097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.2394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.7481 on the scale
276deepseek-v3not in the catalogunstated84.9651
5 columns on 1 board, hardest first are the whole evidence behind 84.96. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.02115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6581 on the scale
277qwen-plus-2025-01-25not in the catalogunstated84.7631
3 columns on 1 board, hardest first are the whole evidence behind 84.76. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6581 on the scale
278microsoft/phi-4unstated84.6431
3 columns on 1 board, hardest first are the whole evidence behind 84.64. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6581 on the scale
279grok-2-1212not in the catalogunstated83.8951
5 columns on 1 board, hardest first are the whole evidence behind 83.89. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6481 on the scale
280gemini-1.5-flash-002not in the catalogunstated83.8651
5 columns on 1 board, hardest first are the whole evidence behind 83.86. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6281 on the scale
281qwen2.5-72b-instructnot in the catalogunstated83.1131
3 columns on 1 board, hardest first are the whole evidence behind 83.11. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6381 on the scale
282llama-4-scout-17b-16e-instructnot in the catalogunstated82.9951
5 columns on 1 board, hardest first are the whole evidence behind 82.99. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.6281 on the scale
283openai/gpt-4o-2024-08-06unstated82.7692
9 columns on 2 boards, hardest first are the whole evidence behind 82.76. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.09119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.26115 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.00115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.00109 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot23.1097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5381 on the scale
284qwen2.5-32b-instructnot in the catalogunstated81.7631
3 columns on 1 board, hardest first are the whole evidence behind 81.76. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5681 on the scale
285claude-3-5-sonnet-20240620not in the catalogunstated81.5971
7 columns on 1 board, hardest first are the whole evidence behind 81.59. Each is a door to the column and everything else scored on it.
- FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub0120 on the scale
- FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub0116 on the scale
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.01115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5281 on the scale
286llama-3.1-405b-instructnot in the catalogunstated81.4931
3 columns on 1 board, hardest first are the whole evidence behind 81.49. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.1094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5081 on the scale
287qwen-turbo-2024-11-01not in the catalogunstated81.3031
3 columns on 1 board, hardest first are the whole evidence behind 81.30. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4283 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5681 on the scale
288mistral-large-2411not in the catalogunstated81.2451
5 columns on 1 board, hardest first are the whole evidence behind 81.24. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.00115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.5183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5081 on the scale
289openai/gpt-4o-2024-05-13unstated81.0131
3 columns on 1 board, hardest first are the whole evidence behind 81.01. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5181 on the scale
290claude-3-5-haiku-20241022not in the catalogunstated80.4262
6 columns on 2 boards, hardest first are the whole evidence behind 80.42. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.00115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot2897 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4681 on the scale
291mistralai/mistral-large-2407unstated80.4031
3 columns on 1 board, hardest first are the whole evidence behind 80.40. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0894 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4581 on the scale
292openai/gpt-4o-2024-11-20unstated80.3772
7 columns on 2 boards, hardest first are the whole evidence behind 80.37. Each is a door to the column and everything else scored on it.
- FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub0.00115 on the scale
- FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub0106 on the scale
- polyglot pass rateAider polyglot18.2097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0694 on the scale
- SWE-Bench verifiedEpoch AI Benchmarking Hub0.3184 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5081 on the scale
293mistral-small-2503not in the catalogunstated80.1831
3 columns on 1 board, hardest first are the whole evidence behind 80.18. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0694 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4781 on the scale
294openai/gpt-4o-mini-2024-07-18unstated80.1082
8 columns on 2 boards, hardest first are the whole evidence behind 80.10. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.12121 on the scale
- Chess PuzzlesEpoch AI Benchmarking Hub0.03119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.08115 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.01109 on the scale
- polyglot pass rateAider polyglot3.6097 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3883 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.5381 on the scale
295gpt-4-turbo-2024-04-09not in the catalogunstated80.0741
4 columns on 1 board, hardest first are the whole evidence behind 80.07. Each is a door to the column and everything else scored on it.
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0.01109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4781 on the scale
296mistral-small-2501not in the catalogunstated79.6231
3 columns on 1 board, hardest first are the whole evidence behind 79.62. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4581 on the scale
297gemini-1.5-pro-001not in the catalogunstated79.2531
3 columns on 1 board, hardest first are the whole evidence behind 79.25. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0794 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4181 on the scale
298meta-llama/llama-3.3-70b-instructunstated79.2231
3 columns on 1 board, hardest first are the whole evidence behind 79.22. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4281 on the scale
299llama-3.1-tulu-3-70b-dponot in the catalogunstated79.2031
3 columns on 1 board, hardest first are the whole evidence behind 79.20. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.4381 on the scale
300claude-3-opus-20240229not in the catalogunstated78.8851
5 columns on 1 board, hardest first are the whole evidence behind 78.88. Each is a door to the column and everything else scored on it.
- Chess PuzzlesEpoch AI Benchmarking Hub0.04119 on the scale
- SimpleQA VerifiedEpoch AI Benchmarking Hub0.13115 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0594 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.3781 on the scale
301llama-3.2-90b-vision-instructnot in the catalogunstated77.9331
3 columns on 1 board, hardest first are the whole evidence behind 77.93. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.3981 on the scale
302meta-llama/llama-3.1-70b-instructunstated77.8631
3 columns on 1 board, hardest first are the whole evidence behind 77.86. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.3781 on the scale
303gemini-1.5-flash-001not in the catalogunstated75.3531
3 columns on 1 board, hardest first are the whole evidence behind 75.35. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4083 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2581 on the scale
304google/gemma-2-27b-itunstated75.0331
3 columns on 1 board, hardest first are the whole evidence behind 75.03. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2881 on the scale
305meta-llama-3-70b-instructnot in the catalogunstated74.9131
3 columns on 1 board, hardest first are the whole evidence behind 74.91. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0494 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2381 on the scale
306mistral-large-2402not in the catalogunstated74.6231
3 columns on 1 board, hardest first are the whole evidence behind 74.62. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3983 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2481 on the scale
307gpt-4-0613not in the catalogunstated74.5641
4 columns on 1 board, hardest first are the whole evidence behind 74.56. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.12121 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2381 on the scale
308hermes-2-theta-llama-3-70bnot in the catalogunstated74.2131
3 columns on 1 board, hardest first are the whole evidence behind 74.21. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2381 on the scale
309claude-3-sonnet-20240229not in the catalogunstated73.4631
3 columns on 1 board, hardest first are the whole evidence behind 73.46. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.4183 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.1881 on the scale
310meta-llama/llama-3.1-8b-instructunstated72.8931
3 columns on 1 board, hardest first are the whole evidence behind 72.89. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2381 on the scale
311gemma-2-9b-itnot in the catalogunstated72.1431
3 columns on 1 board, hardest first are the whole evidence behind 72.14. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.2181 on the scale
312claude-3-haiku-20240307not in the catalogunstated71.8731
3 columns on 1 board, hardest first are the whole evidence behind 71.87. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.1581 on the scale
313claude-2.0not in the catalogunstated70.9131
3 columns on 1 board, hardest first are the whole evidence behind 70.91. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0394 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.1281 on the scale
314gemini-1.0-pro-001not in the catalogunstated70.1831
3 columns on 1 board, hardest first are the whole evidence behind 70.18. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.3483 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.1181 on the scale
315gpt-3.5-turbo-0125not in the catalogunstated69.9251
5 columns on 1 board, hardest first are the whole evidence behind 69.92. Each is a door to the column and everything else scored on it.
- Mystery Game PuzzlesEpoch AI Benchmarking Hub0.03121 on the scale
- FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub0109 on the scale
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2783 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.1281 on the scale
316meta-llama-3-8b-instructnot in the catalogunstated67.0131
3 columns on 1 board, hardest first are the whole evidence behind 67.01. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0294 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.0681 on the scale
317deepseek-llm-67b-chatnot in the catalogunstated66.2331
3 columns on 1 board, hardest first are the whole evidence behind 66.23. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0194 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.0681 on the scale
318llama-2-70b-chat-hfnot in the catalogunstated64.4031
3 columns on 1 board, hardest first are the whole evidence behind 64.40. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.2683 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.0381 on the scale
319mistral-7b-instruct-v0.3not in the catalogunstated60.8431
3 columns on 1 board, hardest first are the whole evidence behind 60.84. Each is a door to the column and everything else scored on it.
- OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub0.0094 on the scale
- GPQA diamondEpoch AI Benchmarking Hub0.1583 on the scale
- MATH level 5Epoch AI Benchmarking Hub0.0481 on the scale
319 of 319 ranked runs, highest on the scale first. Position is the run's place on the scale and stays with it under any order. A row opens onto the cells behind its index, which is the same join the fit was given.
A run on 3 columns and one on 38 print the same width here and are not the same claim, which is why both counts travel on the row. A run whose model the catalog does not list keeps the spelling its publisher used and is marked; it contributes to the fit and to every column difficulty, and carries no price here because nothing in the record prices it. Effort is unstated where the publisher wrote no setting on the run, which is not the same as minimal. The 428 runs scored only on columns kept off this axis are not in the ledger at all, and the column ledger below names those columns and says why.
FrontierMath-ErdosEpoch AI Benchmarking Hub+3.55 · 2.23 · 16 runs
16 runs scored on it, 16 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.10135.84
- openai/gpt-6.1-solmax0.04130.00
- openai/gpt-6-astramax0.03132.41
- google/gemini-4-carbonhigh0.03131.68
- claude-opus-5-5max · not in the catalog0.03128.09
- claude-sonnet-5-5max · not in the catalog0.03122.13
- google/gemini-4-argonhigh0.01124.90
- openai/gpt-6-solmax0.01124.00
- anthropic/claude-fable-5max0122.58
- anthropic/claude-opus-5max0122.24
- google/gemini-3.8-flashhigh0114.21
- moonshotai/kimi-k3max0114.75
- openai/gpt-5.5xhigh0120.29
- openai/gpt-5.6-solmax0123.35
- claude-fable-5-1max · not in the catalog0125.12
- x-ai/grok-4.7xhigh0113.32
SpireClimb (Slay the Spire 1, Ascension 20 Heart)Epoch AI Benchmarking Hub+2.57 · 3.35 · 12 runs
12 runs scored on it, 12 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-6-astramax0.33132.41
- gpt-6.2-astramax · not in the catalog0.33135.84
- openai/gpt-6.1-solmax0.18130.00
- claude-fable-5-2max · not in the catalog0.13132.06
- google/gemini-4-hydrogenxhigh0.10125.83
- openai/gpt-6.1-terramax0.10130.85
- openai/gpt-6-solmax0.05124.00
- claude-opus-5-5max · not in the catalog0.05128.09
- openai/gpt-6-lunamax0113.99
- claude-fable-5-1max · not in the catalog0125.12
- claude-sonnet-5-5max · not in the catalog0122.13
- kimi-k3.1-pro-togethermax · not in the catalog0118.88
SpireClimbEpoch AI Benchmarking Hub+2.48 · 3.35 · 12 runs
12 runs scored on it, 12 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.41135.84
- openai/gpt-6-astramax0.34132.41
- openai/gpt-6.1-solmax0.24130.00
- openai/gpt-6.1-terramax0.22130.85
- claude-fable-5-2max · not in the catalog0.15132.06
- openai/gpt-6-solmax0.09124.00
- google/gemini-4-hydrogenxhigh0.08125.83
- claude-opus-5-5max · not in the catalog0.06128.09
- claude-fable-5-1max · not in the catalog0.02125.12
- claude-sonnet-5-5max · not in the catalog0.01122.13
- openai/gpt-6-lunamax0113.99
- kimi-k3.1-pro-togethermax · not in the catalog0118.88
SpireClimb (Slay the Spire 2, Ascension 10)Epoch AI Benchmarking Hub+2.40 · 3.40 · 12 runs
12 runs scored on it, 12 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.49135.84
- openai/gpt-6-astramax0.35132.41
- openai/gpt-6.1-terramax0.35130.85
- openai/gpt-6.1-solmax0.29130.00
- claude-fable-5-2max · not in the catalog0.17132.06
- openai/gpt-6-solmax0.13124.00
- claude-opus-5-5max · not in the catalog0.07128.09
- google/gemini-4-hydrogenxhigh0.05125.83
- claude-fable-5-1max · not in the catalog0.04125.12
- claude-sonnet-5-5max · not in the catalog0.03122.13
- openai/gpt-6-lunamax0113.99
- kimi-k3.1-pro-togethermax · not in the catalog0118.88
OEIS OpenEpoch AI Benchmarking Hub+1.80 · 1.22 · 3 runs
3 runs scored on it, 1 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-opus-4-8high · not in the catalog0.30set aside
- openai/gpt-5.5medium0.26set aside
- google/gemini-3.5-flashhigh0.22111.12
EBR-benchEpoch AI Benchmarking Hub+1.53 · 2.20 · 34 runs
34 runs scored on it, 33 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.92135.84
- claude-fable-5-2max · not in the catalog0.90132.06
- google/gemini-4-carbonhigh0.88131.68
- openai/gpt-6.1-terramax0.78130.85
- openai/gpt-6-astramax0.76132.41
- claude-opus-5-5max · not in the catalog0.71128.09
- google/gemini-4-hydrogenxhigh0.70125.83
- google/gemini-4-argonhigh0.65124.90
- claude-fable-5-1max · not in the catalog0.57125.12
- openai/gpt-6.1-solmax0.54130.00
- openai/gpt-6-solmax0.53124.00
- anthropic/claude-opus-5max0.46122.24
- openai/gpt-5.6-solmax0.45123.35
- claude-haiku-5-5max · not in the catalog0.44113.88
- anthropic/claude-fable-5max0.40122.58
- kimi-k3.1-pro-togethermax · not in the catalog0.37118.88
- openai/gpt-5.5xhigh0.34120.29
- muse-wonder-1.0max · not in the catalog0.31120.86
- x-ai/grok-4.6xhigh0.30114.39
- claude-opus-4-8max · not in the catalog0.29116.89
- gpt-5.4-2026-03-05xhigh · not in the catalog0.25116.23
- gpt-5.2-2025-12-11xhigh · not in the catalog0.23113.39
- claude-opus-4-7max · not in the catalog0.19113.03
- mistral-large-4-0-osshigh · not in the catalog0.18110.78
- google/gemini-3.1-pro-previewunstated0.14111.66
- claude-opus-4-5-20251101-128kunstated · not in the catalog0.14set aside
- claude-opus-4-6max · not in the catalog0.13111.69
- gpt-5-2025-08-07high · not in the catalog0.13108.61
- qwen/qwen3.7-maxunstated0.10111.06
- z-ai/glm-5.2max0.10110.33
- claude-opus-4-1-20250805unstated · not in the catalog0.0893.81
- google/gemini-3.5-flashhigh0.05111.12
- moonshotai/kimi-k2.6unstated0.02110.82
- claude-sonnet-4-5-20250929unstated · not in the catalog0.0293.21
OEIS Open LiteEpoch AI Benchmarking Hub+1.47 · 1.22 · 5 runs
5 runs scored on it, 2 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5high0.44119.29
- openai/gpt-5.6-solmedium0.43set aside
- claude-opus-4-8high · not in the catalog0.39set aside
- openai/gpt-5.5medium0.36set aside
- google/gemini-3.5-flashhigh0.29111.12
typescriptLiveBench+1.46 · 0.86 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-opus-5-5max · not in the catalog63.33128.09
- deepseek/deepseek-v4.1-flashmax60116.49
- meta/muse-spark-1.3xhigh60116.79
- claude-fable-5-1max · not in the catalog60125.12
- muse-wonder-1.0max · not in the catalog60120.86
- deepseek/deepseek-v4-flash-vision-expunstated56.67116.36
- claude-sonnet-5-5max · not in the catalog56.67122.13
- qwen3.8-maxunstated · not in the catalog56.67118.08
- smaug-agenticunstated · not in the catalog56.67116.30
- anthropic/claude-fable-5max53.33122.58
- moonshotai/kimi-k3unstated53.33116.54
- claude-opus-5-5xhigh · not in the catalog53.33125.14
- smaug-miniunstated · not in the catalog53.33114.73
- anthropic/claude-sonnet-5xhigh50113.50
- meta/muse-spark-1.2xhigh50117.04
- openai/gpt-5.6-solmax50123.35
- qwen/qwen3.8-27bunstated50112.72
- smaug-flashunstated · not in the catalog50116.63
- z-ai/glm-5.3unstated50112.92
- deepseek/deepseek-v4-pro-0813unstated46.67118.60
- google/gemini-3.7-flashhigh46.67116.01
- openai/gpt-5.6-terramax46.67119.24
- openai/gpt-6.1-solxhigh46.67124.64
- claude-sonnet-5-5xhigh · not in the catalog46.67119.88
- qwen3.8-flash-nextunstated · not in the catalog46.67113.12
- z-ai/glm-5.3-flashunstated46.67108.42
- anthropic/claude-opus-5max43.33122.24
- meta/muse-spark-1.1xhigh43.33114.21
- openai/gpt-5.4xhigh43.33119.02
- openai/gpt-5.5xhigh43.33120.29
- openai/gpt-6-astramax43.33132.41
- claude-haiku-5-5max · not in the catalog43.33113.88
- claude-opus-4-7xhigh · not in the catalog43.33113.14
- mistral-large-4high · not in the catalog43.33109.81
- ox-alphamax · not in the catalog43.33106.05
- x-ai/grok-4.6unstated43.33117.62
- x-ai/grok-4.7xhigh43.33113.32
- google/gemini-3.8-flashhigh40114.21
- openai/gpt-6-lunamax40113.99
- openai/gpt-6-solmax40124.00
- openai/gpt-6.1-solmax40130.00
- claude-haiku-5-5xhigh · not in the catalog40112.88
- openai/gpt-5.6-lunamax36.67114.95
- gpt-5.2-2025-12-11high · not in the catalog36.67110.74
- x-ai/grok-4.5unstated36.67114.74
- z-ai/glm-5.2unstated36.67109.40
- google/gemini-3.5-flashhigh33.33111.12
- claude-opus-4-6-thinking-autohigh · not in the catalog33.33113.04
- claude-opus-4-8max · not in the catalog33.33116.89
- x-ai/grok-build-0.1unstated33.33106.25
- thinkingmachines/inklingxhigh30107.98
- deepseek/deepseek-v4-flash-0731unstated26.67113.47
- google/gemini-3.5-flash-litehigh26.6799.97
- google/gemini-3.6-flashhigh26.67110.95
- openai/gpt-5.4-nanoxhigh26.67110.30
- qwen/qwen3.7-maxunstated26.67111.06
- kimi-k2.6-thinkingunstated · not in the catalog26.67107.26
- deepseek/deepseek-v4-flashunstated23.33102.96
- deepseek/deepseek-v4-prounstated23.33111.64
- google/gemini-3.1-pro-previewhigh23.33113.89
- minimax/minimax-m3unstated23.33104.58
- moonshotai/kimi-k2.7-codeunstated23.33107.73
- qwen/qwen3.6-27bunstated23.33103.72
- claude-sonnet-4-6-thinking-automedium · not in the catalog23.33114.06
- openai/gpt-5.2-codexunstated20110.02
- openai/gpt-5.4-minixhigh20104.22
- qwen/qwen3.6-plusunstated20106.49
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog20110.65
- nvidia/nemotron-3-ultra-550b-a55bunstated16.67105.72
- x-ai/grok-4.3unstated13.33103.07
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog6.6787.31
Mystery Game PuzzlesEpoch AI Benchmarking Hub+1.42 · 1.36 · 149 runs
149 runs scored on it, 118 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.95135.84
- claude-fable-5-2max · not in the catalog0.90132.06
- gemini-4-neonxhigh · not in the catalog0.85134.54
- openai/gpt-6-astramax0.84132.41
- openai/gpt-6.1-solmax0.80130.00
- openai/gpt-6.1-terramax0.74130.85
- google/gemini-4-carbonhigh0.73131.68
- claude-opus-5-5max · not in the catalog0.71128.09
- google/gemini-4-hydrogenxhigh0.65125.83
- claude-sonnet-5-5max · not in the catalog0.65122.13
- google/gemini-4-argonhigh0.64124.90
- anthropic/claude-opus-5max0.59122.24
- openai/gpt-5.6-solmax0.58123.35
- claude-fable-5-1max · not in the catalog0.58125.12
- openai/gpt-5.5xhigh0.56120.29
- openai/gpt-6-solmax0.56124.00
- anthropic/claude-fable-5max0.52122.58
- openai/gpt-5.5high0.52set aside
- muse-wonder-1.0max · not in the catalog0.51120.86
- deepseek/deepseek-v4.1-flash-1017max0.49116.25
- google/gemini-3.8-flashhigh0.47114.21
- deepseek/deepseek-v4-pro-0813max0.43113.60
- deepseek/deepseek-v4.1-flashmax0.43116.49
- kimi-k3.1-pro-togethermax · not in the catalog0.42118.88
- qwen3.8-maxxhigh · not in the catalog0.38116.39
- anthropic/claude-opus-5unstated0.37115.94
- google/gemini-3.7-flashhigh0.37116.01
- gpt-5.4-2026-03-05xhigh · not in the catalog0.37116.23
- claude-opus-4-8max · not in the catalog0.36116.89
- anthropic/claude-sonnet-5max0.35112.43
- openai/gpt-5.6-terramax0.35119.24
- deepseek/deepseek-v4-flash-0731max0.34111.14
- google/gemini-3.1-pro-previewhigh0.34113.89
- x-ai/grok-4.6xhigh0.34114.39
- openai/gpt-5.6-solnone0.33100.98
- z-ai/glm-5.3max0.33112.54
- google/gemini-3.1-pro-previewmedium0.32set aside
- google/gemini-3.5-flashhigh0.32111.12
- qwen/qwen3.7-maxunstated0.32111.06
- claude-opus-4-8xhigh · not in the catalog0.31set aside
- google/gemini-3.6-flashhigh0.30110.95
- claude-haiku-5-5max · not in the catalog0.30113.88
- google/gemini-3.1-pro-previewlow0.29set aside
- o3-2025-04-16high · not in the catalog0.29102.96
- x-ai/grok-4.7xhigh0.29113.32
- google/gemini-3.5-flashlow0.28110.83
- openai/gpt-5.5low0.28107.86
- claude-opus-4-7max · not in the catalog0.28113.03
- gpt-5.4-2026-03-05medium · not in the catalog0.28112.04
- google/gemini-3-flash-previewlow0.26set aside
- moonshotai/kimi-k3max0.26114.75
- openai/gpt-5.6-sollow0.26110.85
- qwen/qwen3.7-maxnone0.26set aside
- google/gemini-3-flash-previewminimal0.25set aside
- google/gemini-3.6-flashminimal0.25105.55
- meta/muse-spark-1.3max0.25115.55
- claude-opus-4-6max · not in the catalog0.25111.69
- google/gemini-3.6-flashlow0.23107.80
- gpt-5-2025-08-07high · not in the catalog0.23108.61
- gpt-5.2-2025-12-11high · not in the catalog0.23110.74
- o3-2025-04-16medium · not in the catalog0.23101.11
- qwen/qwen3.6-35b-a3bnone0.22100.22
- claude-opus-4-5-20251101-48kunstated · not in the catalog0.22set aside
- gpt-5.2-2025-12-11medium · not in the catalog0.22108.75
- openai/gpt-5.6-lunamax0.21114.95
- claude-opus-4-1-20250805-24kunstated · not in the catalog0.21set aside
- google/gemini-3-flash-previewhigh0.20113.70
- openai/gpt-5.6-lunanone0.2095.93
- mistral-large-4-0-osshigh · not in the catalog0.20110.78
- nemotron-3-ultraunstated · not in the catalog0.20105.62
- qwen3.5-flashunstated · not in the catalog0.2099.58
- google/gemini-3.5-flash-litelow0.1998.68
- openai/gpt-5.6-terralow0.19106.93
- gpt-5.1-2025-11-13low · not in the catalog0.19100.64
- qwen3.6-max-previewnone · not in the catalog0.19set aside
- z-ai/glm-5.2low0.19105.42
- moonshotai/kimi-k2.6unstated0.18110.82
- openai/gpt-5.5none0.1897.11
- qwen/qwen3.5-397b-a17bnone0.18103.55
- qwen/qwen3.6-flashnone0.18set aside
- z-ai/glm-5.2minimal0.18set aside
- z-ai/glm-5.2none0.1896.22
- deepseek/deepseek-v4-pronone0.1794.28
- mistralai/mistral-large-4-0high0.17110.63
- openai/gpt-5.6-lunalow0.17101.92
- qwen/qwen3.5-122b-a10bnone0.17set aside
- qwen/qwen3.7-plusunstated0.17107.80
- qwen/qwen3.7-plusnone0.17103.17
- claude-sonnet-4-5-20250929-48kunstated · not in the catalog0.17set aside
- gpt-5.4-2026-03-05low · not in the catalog0.17104.16
- qwen3.5-plusnone · not in the catalog0.17set aside
- anthropic/claude-sonnet-5unstated0.16set aside
- claude-sonnet-4-6low · not in the catalog0.16set aside
- gpt-5-2025-08-07low · not in the catalog0.16105.22
- gpt-5.1-2025-11-13medium · not in the catalog0.16105.41
- gpt-5.4-2026-03-05none · not in the catalog0.1696.15
- qwen3.5-flashnone · not in the catalog0.16set aside
- qwen3.5-plusunstated · not in the catalog0.16105.31
- qwen/qwen3.7-flashunstated0.15102.16
- claude-opus-4-6unstated · not in the catalog0.15110.08
- gpt-5-2025-08-07medium · not in the catalog0.15108.12
- gpt-5.1-2025-11-13none · not in the catalog0.1591.90
- z-ai/glm-5.2medium0.15set aside
- meta/muse-spark-1.3xhigh0.14116.79
- openai/gpt-5.6-terranone0.1495.57
- qwen/qwen3.7-flashnone0.14100.54
- claude-sonnet-4-6unstated · not in the catalog0.14set aside
- gpt-5-2025-08-07minimal · not in the catalog0.1495.14
- gpt-5.2-2025-12-11none · not in the catalog0.1496.53
- claude-opus-4-7unstated · not in the catalog0.13set aside
- google/gemini-3.5-flash-liteminimal0.1294.95
- moonshotai/kimi-k2.6none0.12set aside
- moonshotai/kimi-k3low0.12100.12
- openai/gpt-4o-mini-2024-07-18unstated0.1280.10
- openai/gpt-5.6-lunamedium0.12set aside
- qwen/qwen3.6-flashunstated0.12100.95
- qwen/qwen3.6-plusnone0.12set aside
- gpt-4-0613unstated · not in the catalog0.1274.56
- gpt-5.4-mini-2026-03-17none · not in the catalog0.1190.48
- openai/gpt-5.6-terramedium0.10set aside
- gpt-5-mini-2025-08-07minimal · not in the catalog0.1093.86
- gpt-5.2-2025-12-11low · not in the catalog0.10104.89
- qwen/qwen3-235b-a22b-thinking-2507unstated0.09104.10
- gpt-5-nano-2025-08-07low · not in the catalog0.0992.74
- gpt-5-nano-2025-08-07medium · not in the catalog0.0999.03
- gpt-5.4-nano-2026-03-17none · not in the catalog0.0991.33
- minimax/minimax-m3none0.0890.84
- gpt-5-nano-2025-08-07high · not in the catalog0.08100.45
- gpt-5.4-mini-2026-03-17low · not in the catalog0.08set aside
- z-ai/glm-5.3-flashmax0.08108.45
- minimax/minimax-m3unstated0.07104.58
- openai/gpt-6-lunamax0.07113.99
- qwen/qwen3.6-27bnone0.07100.32
- claude-opus-4-6low · not in the catalog0.07set aside
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.0793.48
- gpt-5.4-mini-2026-03-17medium · not in the catalog0.07set aside
- o3-mini-2025-01-31high · not in the catalog0.07102.45
- thinkingmachines/inkling-smallxhigh0.06106.54
- gpt-5.4-nano-2026-03-17medium · not in the catalog0.06set aside
- gpt-5-mini-2025-08-07high · not in the catalog0.05104.57
- gpt-5-nano-2025-08-07minimal · not in the catalog0.0588.08
- gpt-5.4-nano-2026-03-17high · not in the catalog0.05106.76
- o4-mini-2025-04-16high · not in the catalog0.05103.35
- qwen3-max-2025-09-23unstated · not in the catalog0.05100.03
- gpt-5-mini-2025-08-07medium · not in the catalog0.04103.83
- gpt-3.5-turbo-0125unstated · not in the catalog0.0369.92
- gpt-5.4-nano-2026-03-17low · not in the catalog0.0398.40
- openai/gpt-oss-120bmedium0.02set aside
- openai/gpt-oss-120bhigh0set aside
FrontierMath-Tier-4-2025-07-01-PrivateEpoch AI Benchmarking Hub+1.36 · 2.81 · 72 runs
72 runs scored on it, 63 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gdm-ai-co-mathematicianunstated · not in the catalog0.48set aside
- gpt-5.5-pro-pre-releasehigh · not in the catalog0.40set aside
- gpt-5.5-pro-pre-releasexhigh · not in the catalog0.40117.38
- gpt-5.4-pro-2026-03-05-web-appunstated · not in the catalog0.38set aside
- gpt-5.5-pre-releasexhigh · not in the catalog0.35116.86
- gpt-5.2-pro-2025-12-11-webappunstated · not in the catalog0.31set aside
- claude-opus-4-8max · not in the catalog0.31116.89
- gpt-5.4-2026-03-05xhigh · not in the catalog0.27116.23
- claude-opus-4-7xhigh · not in the catalog0.23113.14
- claude-opus-4-6max · not in the catalog0.23111.69
- claude-opus-4-6-32kunstated · not in the catalog0.21111.03
- claude-opus-4-6-64kunstated · not in the catalog0.21111.92
- gpt-5.2-2025-12-11high · not in the catalog0.19110.74
- gpt-5.2-2025-12-11xhigh · not in the catalog0.19113.39
- gemini-3-pro-previewunstated · not in the catalog0.19110.74
- google/gemini-3.1-pro-previewunstated0.17111.66
- gpt-5.2-2025-12-11medium · not in the catalog0.17108.75
- gpt-5-pro-2025-10-06high · not in the catalog0.15109.23
- muse-sparkunstated · not in the catalog0.15110.41
- google/gemini-3.5-flashhigh0.15111.12
- claude-opus-4-6unstated · not in the catalog0.15110.08
- moonshotai/kimi-k2.6unstated0.15110.82
- gpt-5-2025-08-07high · not in the catalog0.13108.61
- gpt-5.1-2025-11-13high · not in the catalog0.13109.94
- z-ai/glm-5.1unstated0.13109.27
- gemini-2.5-deep-think-2025-08-01-webappunstated · not in the catalog0.10set aside
- qwen/qwen3.6-plusunstated0.08106.49
- claude-sonnet-4-6-16kunstated · not in the catalog0.08109.98
- gpt-5-2025-08-07medium · not in the catalog0.06108.12
- gpt-5-mini-2025-08-07high · not in the catalog0.06104.57
- gpt-5.2-2025-12-11low · not in the catalog0.06104.89
- gpt-5.4-nano-2026-03-17high · not in the catalog0.06106.76
- o4-mini-2025-04-16high · not in the catalog0.06103.35
- fireworks/kimi-k2p5unstated · not in the catalog0.04107.32
- google/gemini-2.5-flashunstated0.04set aside
- google/gemini-2.5-prounstated0.04103.73
- google/gemini-3-flash-previewunstated0.04108.16
- claude-opus-4-1-20250805-27kunstated · not in the catalog0.0497.29
- claude-opus-4-20250514-27kunstated · not in the catalog0.0497.69
- claude-opus-4-5-20251101unstated · not in the catalog0.0499.92
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.04104.05
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.04101.18
- gpt-5-mini-2025-08-07medium · not in the catalog0.04103.83
- gpt-5.1-2025-11-13medium · not in the catalog0.04105.41
- o3-mini-2025-01-31high · not in the catalog0.04102.45
- qwen3.6-max-previewunstated · not in the catalog0.04106.75
- zai-org/glm-4.6unstated · not in the catalog0.0292.21
- fireworks/deepseek-v3p2unstated · not in the catalog0.02set aside
- z-ai/glm-5unstated0.0299.96
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.0296.01
- claude-opus-4-5-20251101-16kunstated · not in the catalog0.02104.70
- claude-sonnet-4-5-20250929unstated · not in the catalog0.0293.21
- gemini-2.5-pro-preview-06-05unstated · not in the catalog0.02103.27
- gpt-5-nano-2025-08-07medium · not in the catalog0.0299.03
- grok-4-0709unstated · not in the catalog0.02104.16
- o3-2025-04-16high · not in the catalog0.02102.96
- o4-mini-2025-04-16medium · not in the catalog0.02101.93
- qwen3.5-plusunstated · not in the catalog0.02105.31
- gpt-5.4-mini-2026-03-17high · not in the catalog0.02106.09
- grok-4-heavy-web-appunstated · not in the catalog0.02set aside
- qwen/qwen3.6-flashunstated0100.95
- claude-3-5-sonnet-20240620unstated · not in the catalog081.59
- claude-3-5-sonnet-20241022unstated · not in the catalog085.31
- claude-opus-4-20250514unstated · not in the catalog092.48
- claude-sonnet-4-20250514unstated · not in the catalog090.16
- gpt-4.1-2025-04-14unstated · not in the catalog091.64
- gpt-5-nano-2025-08-07high · not in the catalog0100.45
- grok-3-betaunstated · not in the catalog094.65
- moonshotai/kimi-k2-thinkingunstated · not in the catalog0104.82
- qwen/qwen3-235b-a22b-thinking-2507unstated · not in the catalog0101.29
- qwen3.5-flashunstated · not in the catalog099.58
- zai-org/glm-4.7unstated · not in the catalog0set aside
Furniture AssemblyEpoch AI Benchmarking Hub+1.31 · 1.89 · 42 runs
42 runs scored on it, 41 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-fable-5-2max · not in the catalog0.90132.06
- gemini-4-neonxhigh · not in the catalog0.89134.54
- gpt-6.2-astramax · not in the catalog0.89135.84
- claude-opus-5-5max · not in the catalog0.83128.09
- openai/gpt-6-astramax0.80132.41
- openai/gpt-6.1-solmax0.80130.00
- openai/gpt-6.1-terramax0.79130.85
- google/gemini-4-carbonhigh0.78131.68
- claude-sonnet-5-5max · not in the catalog0.75122.13
- google/gemini-4-argonhigh0.72124.90
- claude-fable-5-1max · not in the catalog0.70125.12
- anthropic/claude-opus-5max0.61122.24
- openai/gpt-6-solmax0.58124.00
- google/gemini-4-hydrogenxhigh0.57125.83
- openai/gpt-5.6-solmax0.57123.35
- kimi-k3.1-pro-togethermax · not in the catalog0.56118.88
- openai/gpt-5.6-terramax0.54119.24
- muse-wonder-1.0max · not in the catalog0.54120.86
- claude-haiku-5-5max · not in the catalog0.47113.88
- openai/gpt-5.5xhigh0.44120.29
- openai/gpt-6-lunamax0.44113.99
- openai/gpt-5.6-lunamax0.42114.95
- claude-opus-4-8max · not in the catalog0.42116.89
- x-ai/grok-4.6xhigh0.40114.39
- gpt-5.2-2025-12-11xhigh · not in the catalog0.38113.39
- gpt-5.4-2026-03-05xhigh · not in the catalog0.38116.23
- anthropic/claude-fable-5max0.36122.58
- deepseek/deepseek-v4.1-flashmax0.34116.49
- moonshotai/kimi-k3max0.34114.75
- claude-opus-4-7max · not in the catalog0.33113.03
- google/gemini-3.8-flashhigh0.32114.21
- mistral-large-4-0-osshigh · not in the catalog0.31110.78
- claude-opus-4-5-20251101-64kunstated · not in the catalog0.28set aside
- claude-opus-4-6max · not in the catalog0.28111.69
- google/gemini-3.1-pro-previewhigh0.27113.89
- google/gemini-3.7-flashhigh0.27116.01
- google/gemini-3.6-flashhigh0.23110.95
- x-ai/grok-4.5high0.23111.55
- moonshotai/kimi-k2.6unstated0.22110.82
- mistralai/mistral-large-4-0high0.21110.63
- x-ai/grok-4.7xhigh0.21113.32
- qwen/qwen3.8-max-0902xhigh0.20113.51
Chess PuzzlesEpoch AI Benchmarking Hub+1.26 · 1.39 · 167 runs
167 runs scored on it, 159 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gemini-4-neonxhigh · not in the catalog0.83134.54
- claude-fable-5-2max · not in the catalog0.79132.06
- google/gemini-4-carbonhigh0.78131.68
- gpt-6.2-astramax · not in the catalog0.78135.84
- gpt-6-astra-promaxunstated · not in the catalog0.76set aside
- openai/gpt-6-astramax0.74132.41
- openai/gpt-6.1-solmax0.68130.00
- gpt-5.6-sol-promaxunstated · not in the catalog0.68set aside
- claude-opus-5-5max · not in the catalog0.67128.09
- google/gemini-3.7-flashhigh0.66116.01
- openai/gpt-5.5xhigh0.63120.29
- openai/gpt-5.5-proxhigh0.62122.66
- google/gemini-4-hydrogenxhigh0.60125.83
- gpt-5.2-pro-2025-12-11xhigh · not in the catalog0.60116.74
- google/gemini-4-argonhigh0.59124.90
- openai/gpt-5.6-solmax0.59123.35
- claude-sonnet-5-5max · not in the catalog0.59122.13
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.59117.92
- google/gemini-3.8-flashhigh0.58114.21
- openai/gpt-6-solmax0.56124.00
- claude-haiku-5-5max · not in the catalog0.55113.88
- claude-opus-4-8max · not in the catalog0.52116.89
- google/gemini-3.1-pro-previewhigh0.51113.89
- deepseek/deepseek-v4.1-flashmax0.48116.49
- claude-fable-5-1max · not in the catalog0.48125.12
- gpt-5-pro-2025-10-06high · not in the catalog0.48109.23
- gpt-5.2-2025-12-11xhigh · not in the catalog0.48113.39
- gpt-5.4-2026-03-05xhigh · not in the catalog0.48116.23
- anthropic/claude-fable-5max0.47122.58
- openai/gpt-5.6-terramax0.47119.24
- deepseek/deepseek-v4-pro-0813max0.46113.60
- deepseek/deepseek-v4.1-flash-1017max0.46116.25
- qwen3.8-maxxhigh · not in the catalog0.46116.39
- anthropic/claude-fable-5high0.45119.29
- anthropic/claude-sonnet-5max0.45112.43
- google/gemini-3.5-flashhigh0.45111.12
- gpt-5.2-2025-12-11high · not in the catalog0.45110.74
- gpt-5.4-nano-2026-03-17xhigh · not in the catalog0.45set aside
- kimi-k3.1-pro-togethermax · not in the catalog0.45118.88
- anthropic/claude-opus-5max0.43122.24
- openai/gpt-5.6-lunamax0.43114.95
- google/gemini-3.5-flashlow0.42110.83
- google/gemini-3.6-flashlow0.42107.80
- gpt-5.1-2025-11-13high · not in the catalog0.42109.94
- gpt-5.4-2026-03-05high · not in the catalog0.42115.98
- claude-opus-4-7max · not in the catalog0.41113.03
- x-ai/grok-4.7xhigh0.41113.32
- google/gemini-3-flash-previewhigh0.40113.70
- google/gemini-3.6-flashhigh0.40110.95
- doubao-seed-evolvinghigh · not in the catalog0.40114.76
- meta/muse-spark-1.3max0.37115.55
- moonshotai/kimi-k3high0.37112.20
- qwen/qwen3.8-max-0902xhigh0.37113.51
- gpt-5.4-2026-03-05medium · not in the catalog0.37112.04
- o3-2025-04-16medium · not in the catalog0.37101.11
- z-ai/glm-5.3max0.37112.54
- moonshotai/kimi-k3max0.36114.75
- openai/gpt-6-lunamax0.36113.99
- claude-opus-4-7xhigh · not in the catalog0.36113.14
- gpt-5-2025-08-07high · not in the catalog0.36108.61
- gpt-5.2-2025-12-11medium · not in the catalog0.36108.75
- gpt-5.4-nano-2026-03-17high · not in the catalog0.36106.76
- o3-2025-04-16high · not in the catalog0.36102.96
- o3-2025-04-16low · not in the catalog0.3699.37
- x-ai/grok-4.5high0.35111.55
- anthropic/claude-sonnet-5xhigh0.34113.50
- meta/muse-spark-1.1xhigh0.34114.21
- meta/muse-spark-1.3xhigh0.34116.79
- mistralai/mistral-large-4-0high0.33110.63
- x-ai/grok-4.6high0.33114.02
- x-ai/grok-4.6xhigh0.33114.39
- deepseek/deepseek-v4-flash-0731max0.32111.14
- gpt-5-2025-08-07medium · not in the catalog0.32108.12
- anthropic/claude-fable-5low0.31110.20
- openai/gpt-5.6-sollow0.31110.85
- openai/gpt-5.5low0.30107.86
- google/gemini-3.6-flashminimal0.29105.55
- claude-opus-4-8low · not in the catalog0.29111.54
- grok-4.20-0309-reasoningunstated · not in the catalog0.29108.33
- mistral-large-4-0-osshigh · not in the catalog0.29110.78
- anthropic/claude-opus-5low0.27109.13
- google/gemini-3.5-flashminimal0.27104.41
- meta/muse-spark-1.2xhigh0.27117.04
- openai/gpt-5.6-terralow0.27106.93
- qwen/qwen3.7-maxunstated0.27111.06
- gpt-5-2025-08-07low · not in the catalog0.27105.22
- claude-opus-4-7low · not in the catalog0.26set aside
- doubao-seed-2.1-turbo-260628high · not in the catalog0.26108.84
- gpt-5.2-2025-12-11low · not in the catalog0.26104.89
- claude-haiku-5-5low · not in the catalog0.24set aside
- gpt-5-mini-2025-08-07high · not in the catalog0.24104.57
- claude-sonnet-4-6max · not in the catalog0.23101.17
- gpt-5.1-2025-11-13low · not in the catalog0.23100.64
- gpt-5.4-mini-2026-03-17high · not in the catalog0.23106.09
- gpt-5-nano-2025-08-07high · not in the catalog0.22100.45
- o4-mini-2025-04-16medium · not in the catalog0.22101.93
- z-ai/glm-5.2max0.22110.33
- google/gemini-3.5-flash-litelow0.2198.68
- thinkingmachines/inklingxhigh0.21107.98
- claude-opus-4-6max · not in the catalog0.21111.69
- doubao-seed-2.1-pro-260628high · not in the catalog0.21109.61
- z-ai/glm-4.7unstated0.21104.07
- z-ai/glm-5.3-flashmax0.21108.45
- moonshotai/kimi-k2.6unstated0.20110.82
- doubao-seed-2.0-lite-260428high · not in the catalog0.20106.42
- doubao-seed-2.0-pro-260215high · not in the catalog0.20108.12
- gpt-5.4-nano-2026-03-17low · not in the catalog0.2098.40
- z-ai/glm-5.1unstated0.20109.27
- deepseek/deepseek-v4-promax0.19107.02
- claude-opus-4-8none · not in the catalog0.19104.31
- claude-sonnet-4-6high · not in the catalog0.19102.30
- gpt-5.4-2026-03-05low · not in the catalog0.19104.16
- z-ai/glm-5.2none0.1996.22
- moonshotai/kimi-k3low0.18100.12
- x-ai/grok-4.3high0.18107.39
- z-ai/glm-5.2low0.18105.42
- minimax/minimax-m3unstated0.17104.58
- openai/gpt-5.6-lunalow0.17101.92
- gpt-5.4-mini-2026-03-17xhigh · not in the catalog0.17106.91
- deepseek/deepseek-v4-prohigh0.16108.20
- google/gemini-3.5-flash-litehigh0.1699.97
- qwen/qwen3.6-plusunstated0.16106.49
- claude-sonnet-4-6medium · not in the catalog0.16103.01
- gpt-5-mini-2025-08-07low · not in the catalog0.16set aside
- gpt-5-nano-2025-08-07low · not in the catalog0.1592.74
- o3-mini-2025-01-31high · not in the catalog0.15102.45
- o4-mini-2025-04-16low · not in the catalog0.15100.61
- moonshotai/kimi-k2.7-codeunstated0.14107.73
- claude-opus-4-5-20251101-59kunstated · not in the catalog0.14set aside
- qwen3.5-flashunstated · not in the catalog0.1499.58
- o4-mini-2025-04-16high · not in the catalog0.13103.35
- qwen/qwen3.6-flashunstated0.12100.95
- gpt-4.1-2025-04-14unstated · not in the catalog0.1191.64
- gpt-5-2025-08-07minimal · not in the catalog0.1195.14
- o1-2024-12-17low · not in the catalog0.1194.71
- o3-mini-2025-01-31medium · not in the catalog0.1197.94
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.10101.18
- claude-sonnet-4-5-20250929-59kunstated · not in the catalog0.10101.26
- o1-2024-12-17high · not in the catalog0.1098.27
- openai/gpt-4o-2024-08-06unstated0.0982.76
- openai/gpt-5.5none0.0997.11
- qwen3.5-plusunstated · not in the catalog0.09105.31
- minimax/minimax-m3none0.0890.84
- claude-haiku-4-5-20251001-59kunstated · not in the catalog0.08set aside
- o1-2024-12-17medium · not in the catalog0.0897.81
- google/gemini-3.5-flash-liteminimal0.0794.95
- openai/gpt-5.6-solnone0.07100.98
- claude-sonnet-4-5-20250929unstated · not in the catalog0.0693.21
- gpt-5.1-2025-11-13none · not in the catalog0.0691.90
- o3-mini-2025-01-31low · not in the catalog0.0691.91
- openai/gpt-5.6-terranone0.0595.57
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.0596.01
- claude-opus-4-5-20251101unstated · not in the catalog0.0599.92
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.05104.05
- seed-1.6unstated · not in the catalog0.0597.70
- deepseek/deepseek-v4-pronone0.0494.28
- claude-3-opus-20240229unstated · not in the catalog0.0478.88
- openai/gpt-4o-mini-2024-07-18unstated0.0380.10
- openai/gpt-5.6-lunanone0.0395.93
- gpt-5.4-2026-03-05none · not in the catalog0.0396.15
- gpt-5-nano-2025-08-07minimal · not in the catalog0.0288.08
- gpt-5.2-2025-12-11none · not in the catalog0.0296.53
- gpt-5.4-mini-2026-03-17none · not in the catalog0.0290.48
- gpt-5.4-nano-2026-03-17none · not in the catalog0.0291.33
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.0193.48
- gpt-4.1-nano-2025-04-14unstated · not in the catalog086.22
- gpt-5-mini-2025-08-07minimal · not in the catalog093.86
MirrorCodeEpoch AI Benchmarking Hub+1.24 · 1.35 · 48 runs
48 runs scored on it, 42 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-sonnet-5-5max · not in the catalog0.98122.13
- claude-opus-5-5max · not in the catalog0.96128.09
- claude-fable-5-1max · not in the catalog0.93125.12
- anthropic/claude-opus-5max0.88122.24
- claude-opus-5-5high · not in the catalog0.79set aside
- claude-haiku-5-5max · not in the catalog0.79113.88
- openai/gpt-6.1-terramax0.77130.85
- claude-fable-5-1high · not in the catalog0.73set aside
- anthropic/claude-fable-5max0.70122.58
- gpt-6.2-astramax · not in the catalog0.68135.84
- claude-opus-4-8max · not in the catalog0.66116.89
- deepseek/deepseek-v4.1-flashmax0.64116.49
- anthropic/claude-fable-5high0.64119.29
- anthropic/claude-sonnet-5max0.63112.43
- openai/gpt-6.1-solmax0.61130.00
- openai/gpt-6-astramax0.60132.41
- google/gemini-4-argonhigh0.59124.90
- meta/muse-spark-1.3max0.54115.55
- moonshotai/kimi-k3max0.54114.75
- x-ai/grok-4.7xhigh0.54113.32
- qwen3.8-maxxhigh · not in the catalog0.52116.39
- openai/gpt-6-solmax0.49124.00
- qwen/qwen3.8-max-0902xhigh0.48113.51
- openai/gpt-6-astrahigh0.47set aside
- muse-wonder-1.0max · not in the catalog0.46120.86
- claude-opus-4-7max · not in the catalog0.43113.03
- kimi-k3.1-pro-togethermax · not in the catalog0.43118.88
- google/gemini-3.7-flashhigh0.41116.01
- openai/gpt-5.6-terramax0.41119.24
- x-ai/grok-4.6xhigh0.41114.39
- mistralai/mistral-large-4-0high0.40110.63
- google/gemini-3.8-flashhigh0.39114.21
- z-ai/glm-5.2max0.37110.33
- deepseek/deepseek-v4-pro-0813max0.33113.60
- claude-opus-4-6max · not in the catalog0.33111.69
- gpt-5.2-2025-12-11xhigh · not in the catalog0.32113.39
- x-ai/grok-4.5high0.32111.55
- claude-opus-4-7high · not in the catalog0.31set aside
- openai/gpt-6-lunamax0.27113.99
- google/gemini-3.6-flashhigh0.23110.95
- openai/gpt-5.5xhigh0.23120.29
- gpt-5.4-2026-03-05xhigh · not in the catalog0.23116.23
- openai/gpt-5.6-lunamax0.21114.95
- openai/gpt-5.6-solhigh0.20set aside
- gpt-5.4-2026-03-05high · not in the catalog0.16115.98
- openai/gpt-5.5high0.10set aside
- google/gemini-3.1-pro-previewhigh0.09113.89
- gpt-5-2025-08-07high · not in the catalog0.04108.61
tablejoinLiveBench+1.16 · 0.32 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-6-astramax58.37132.41
- openai/gpt-6.1-solmax57.42130.00
- claude-fable-5-1max · not in the catalog56.15125.12
- anthropic/claude-fable-5max56.13122.58
- openai/gpt-5.5xhigh55.90120.29
- openai/gpt-6.1-solxhigh55.25124.64
- openai/gpt-6-solmax54.02124.00
- google/gemini-3.1-pro-previewhigh52.35113.89
- claude-opus-5-5max · not in the catalog52.33128.09
- anthropic/claude-opus-5max51.96122.24
- smaug-miniunstated · not in the catalog51.87114.73
- openai/gpt-5.4xhigh51.75119.02
- smaug-agenticunstated · not in the catalog51.23116.30
- claude-opus-4-8max · not in the catalog51.08116.89
- qwen3.8-flash-nextunstated · not in the catalog51.02113.12
- claude-opus-5-5xhigh · not in the catalog51125.14
- smaug-flashunstated · not in the catalog50.71116.63
- openai/gpt-5.4-nanoxhigh50.59110.30
- google/gemini-3.5-flashhigh50.50111.12
- deepseek/deepseek-v4-flash-vision-expunstated50.35116.36
- openai/gpt-5.6-terramax50.25119.24
- qwen3.8-maxunstated · not in the catalog50.06118.08
- openai/gpt-6-lunamax49.62113.99
- deepseek/deepseek-v4-pro-0813unstated49.46118.60
- openai/gpt-5.4-minixhigh49.38104.22
- openai/gpt-5.6-solmax49.31123.35
- claude-haiku-5-5xhigh · not in the catalog49.17112.88
- mistral-large-4high · not in the catalog49.10109.81
- deepseek/deepseek-v4.1-flashmax49.04116.49
- z-ai/glm-5.3-flashunstated48.83108.42
- qwen/qwen3.8-27bunstated48.75112.72
- deepseek/deepseek-v4-flash-0731unstated48.54113.47
- thinkingmachines/inklingxhigh48.48107.98
- moonshotai/kimi-k3unstated48.40116.54
- deepseek/deepseek-v4-prounstated48.19111.64
- meta/muse-spark-1.3xhigh48.19116.79
- claude-opus-4-6-thinking-autohigh · not in the catalog48.17113.04
- openai/gpt-5.6-lunamax47.60114.95
- moonshotai/kimi-k2.7-codeunstated47.54107.73
- claude-opus-4-7xhigh · not in the catalog47.23113.14
- google/gemini-3.6-flashhigh47.13110.95
- ox-alphamax · not in the catalog46.94106.05
- claude-sonnet-5-5xhigh · not in the catalog46.88119.88
- meta/muse-spark-1.2xhigh46.83117.04
- claude-haiku-5-5max · not in the catalog46.67113.88
- meta/muse-spark-1.1xhigh46.56114.21
- nvidia/nemotron-3-ultra-550b-a55bunstated46.56105.72
- muse-wonder-1.0max · not in the catalog46.33120.86
- deepseek/deepseek-v4-flashunstated46.12102.96
- kimi-k2.6-thinkingunstated · not in the catalog46.10107.26
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog45.92110.65
- z-ai/glm-5.2unstated45.87109.40
- minimax/minimax-m3unstated45.85104.58
- openai/gpt-5.2-codexunstated45.60110.02
- gpt-5.2-2025-12-11high · not in the catalog45.58110.74
- qwen/qwen3.7-maxunstated45.52111.06
- google/gemini-3.7-flashhigh45.48116.01
- google/gemini-3.8-flashhigh45.33114.21
- claude-sonnet-4-6-thinking-automedium · not in the catalog45.08114.06
- x-ai/grok-4.7xhigh44.92113.32
- qwen/qwen3.6-plusunstated44.65106.49
- x-ai/grok-4.5unstated43.58114.74
- x-ai/grok-4.6unstated43.42117.62
- qwen/qwen3.6-27bunstated42.85103.72
- x-ai/grok-build-0.1unstated42.65106.25
- anthropic/claude-sonnet-5xhigh42.44113.50
- z-ai/glm-5.3unstated42.35112.92
- x-ai/grok-4.3unstated41.25103.07
- google/gemini-3.5-flash-litehigh40.6299.97
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog40.4887.31
- claude-sonnet-5-5max · not in the catalog39.81122.13
FrontierMath-Tier-4-2025-07-01-PublicEpoch AI Benchmarking Hub+1.10 · 11.86 · 48 runs
48 runs scored on it, 47 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5max1122.58
- gpt-5.4-2026-03-05high · not in the catalog0.50115.98
- gpt-5.4-2026-03-05xhigh · not in the catalog0.50116.23
- deepseek/deepseek-r1-0528unstated098.81
- google/gemini-2.5-prounstated0103.73
- google/gemini-3-flash-previewunstated0108.16
- google/gemini-3.1-pro-previewunstated0111.66
- google/gemini-3.5-flashhigh0111.12
- qwen/qwen3.6-plusunstated0106.49
- claude-3-5-sonnet-20240620unstated · not in the catalog081.59
- claude-3-5-sonnet-20241022unstated · not in the catalog085.31
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog095.18
- claude-opus-4-20250514unstated · not in the catalog092.48
- claude-opus-4-20250514-27kunstated · not in the catalog097.69
- claude-opus-4-6max · not in the catalog0111.69
- claude-opus-4-6-32kunstated · not in the catalog0111.03
- claude-opus-4-6-64kunstated · not in the catalog0111.92
- claude-opus-4-7xhigh · not in the catalog0113.14
- claude-sonnet-4-20250514unstated · not in the catalog090.16
- claude-sonnet-4-20250514-59kunstated · not in the catalog099.17
- claude-sonnet-4-5-20250929unstated · not in the catalog093.21
- claude-sonnet-4-6-16kunstated · not in the catalog0109.98
- gemini-2.5-pro-preview-06-05unstated · not in the catalog0103.27
- gemini-3-pro-previewunstated · not in the catalog0110.74
- gpt-4.1-2025-04-14unstated · not in the catalog091.64
- gpt-5-2025-08-07high · not in the catalog0108.61
- gpt-5-2025-08-07medium · not in the catalog0108.12
- gpt-5-mini-2025-08-07high · not in the catalog0104.57
- gpt-5-mini-2025-08-07medium · not in the catalog0103.83
- gpt-5-nano-2025-08-07high · not in the catalog0100.45
- gpt-5-nano-2025-08-07medium · not in the catalog099.03
- gpt-5.2-2025-12-11high · not in the catalog0110.74
- gpt-5.2-2025-12-11low · not in the catalog0104.89
- gpt-5.2-2025-12-11medium · not in the catalog0108.75
- grok-3-betaunstated · not in the catalog094.65
- grok-3-mini-betahigh · not in the catalog097.03
- grok-4-0709unstated · not in the catalog0104.16
- moonshotai/kimi-k2-thinkingunstated · not in the catalog0104.82
- o3-2025-04-16high · not in the catalog0102.96
- o3-mini-2025-01-31high · not in the catalog0102.45
- o4-mini-2025-04-16high · not in the catalog0103.35
- o4-mini-2025-04-16medium · not in the catalog0101.93
- qwen/qwen3-235b-a22b-thinking-2507unstated · not in the catalog0101.29
- qwen3.6-max-previewunstated · not in the catalog0106.75
- zai-org/glm-4.6unstated · not in the catalog092.21
- z-ai/glm-4.5unstated0set aside
- z-ai/glm-5unstated099.96
- z-ai/glm-5.1unstated0109.27
FrontierMath-Tier-4-v2-PrivateEpoch AI Benchmarking Hub+1.09 · 3.33 · 84 runs
84 runs scored on it, 76 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-6.1-solmax1130.00
- claude-fable-5-2max · not in the catalog1132.06
- gpt-6.2-astramax · not in the catalog1135.84
- openai/gpt-6-astrahigh0.98set aside
- openai/gpt-6-astramax0.98132.41
- openai/gpt-6-astraxhigh0.98set aside
- openai/gpt-6-astramedium0.98set aside
- gemini-4-neonxhigh · not in the catalog0.98134.54
- google/gemini-4-carbonhigh0.95131.68
- claude-opus-5-5max · not in the catalog0.95128.09
- openai/gpt-6.1-terramax0.93130.85
- anthropic/claude-fable-5max0.90122.58
- openai/gpt-6-solmax0.90124.00
- google/gemini-4-hydrogenxhigh0.88125.83
- openai/gpt-6-astralow0.88set aside
- claude-fable-5-1max · not in the catalog0.88125.12
- openai/gpt-5.6-solmax0.83123.35
- openai/gpt-6-astranone0.83set aside
- claude-sonnet-5-5max · not in the catalog0.80122.13
- gpt-5.6-sol-promaxunstated · not in the catalog0.80set aside
- google/gemini-4-argonhigh0.78124.90
- openai/gpt-5.5-proxhigh0.78122.66
- gdm-ai-co-mathematicianunstated · not in the catalog0.76set aside
- anthropic/claude-opus-5max0.73122.24
- openai/gpt-5.5xhigh0.72120.29
- openai/gpt-5.6-terramax0.71119.24
- muse-wonder-1.0max · not in the catalog0.71120.86
- kimi-k3.1-pro-togethermax · not in the catalog0.66118.88
- openai/gpt-5.6-lunamax0.61114.95
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.59117.92
- openai/gpt-6-lunamax0.56113.99
- claude-opus-4-8max · not in the catalog0.56116.89
- gpt-5.4-2026-03-05xhigh · not in the catalog0.49116.23
- deepseek/deepseek-v4.1-flash-1017max0.46116.25
- meta/muse-spark-1.3max0.46115.55
- claude-haiku-5-5max · not in the catalog0.46113.88
- qwen3.8-maxxhigh · not in the catalog0.46116.39
- gpt-5.2-pro-2025-12-11xhigh · not in the catalog0.46116.74
- meta/muse-spark-1.3xhigh0.41116.79
- moonshotai/kimi-k3max0.39114.75
- google/gemini-3.7-flashhigh0.37116.01
- qwen/qwen3.7-maxunstated0.34111.06
- qwen/qwen3.8-max-0902xhigh0.34113.51
- claude-opus-4-7max · not in the catalog0.32113.03
- muse-spark-open-816b-a42bxhigh · not in the catalog0.32113.92
- x-ai/grok-4.6xhigh0.32114.39
- gpt-5.2-2025-12-11xhigh · not in the catalog0.32113.39
- anthropic/claude-sonnet-5max0.29112.43
- z-ai/glm-5.2max0.29110.33
- z-ai/glm-5.3max0.29112.54
- deepseek/deepseek-v4-pro-0813max0.27113.60
- deepseek/deepseek-v4.1-flashmax0.27116.49
- google/gemini-3.1-pro-previewunstated0.27111.66
- google/gemini-3.5-flashhigh0.27111.12
- mistralai/mistral-large-4-0high0.27110.63
- claude-opus-4-6max · not in the catalog0.27111.69
- moonshotai/kimi-k2.6unstated0.26110.82
- deepseek/deepseek-v4-flash-0731max0.24111.14
- x-ai/grok-4.5high0.24111.55
- google/gemini-3.6-flashhigh0.22110.95
- google/gemini-3.8-flashhigh0.22114.21
- gpt-5-2025-08-07high · not in the catalog0.22108.61
- gpt-5-pro-2025-10-06high · not in the catalog0.20109.23
- google/gemini-3-flash-previewunstated0.17108.16
- thinkingmachines/inkling-smallxhigh0.17106.54
- grok-4.20-0309-reasoningunstated · not in the catalog0.17108.33
- x-ai/grok-4.7xhigh0.17113.32
- z-ai/glm-5.3-flashmax0.17108.45
- x-ai/grok-4.3high0.15107.39
- moonshotai/kimi-k2.7-codeunstated0.12107.73
- gpt-5-mini-2025-08-07high · not in the catalog0.12104.57
- gpt-5.4-nano-2026-03-17high · not in the catalog0.12106.76
- gpt-5.4-mini-2026-03-17xhigh · not in the catalog0.10106.91
- thinkingmachines/inklingxhigh0.05107.98
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.05104.05
- o4-mini-2025-04-16high · not in the catalog0.05103.35
- deepseek/deepseek-v4-promax0.02107.02
- claude-opus-4-1-20250805-32kunstated · not in the catalog0.02set aside
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.02101.18
- gpt-5-nano-2025-08-07high · not in the catalog0.02100.45
- gpt-5.5-instantunstated · not in the catalog0.0299.89
- google/gemini-2.5-prounstated0103.73
- google/gemini-3.5-flash-litehigh099.97
- o3-mini-2025-01-31high · not in the catalog0102.45
SimpleQA VerifiedEpoch AI Benchmarking Hub+1.03 · 0.99 · 106 runs
106 runs scored on it, 102 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-fable-5-2max · not in the catalog0.94132.06
- gpt-6.2-astramax · not in the catalog0.79135.84
- google/gemini-4-carbonhigh0.78131.68
- google/gemini-4-argonhigh0.76124.90
- openai/gpt-6-astramax0.76132.41
- gemini-4-neonxhigh · not in the catalog0.75134.54
- openai/gpt-6.1-solmax0.74130.00
- google/gemini-3.1-pro-previewhigh0.73113.89
- claude-opus-5-5max · not in the catalog0.72128.09
- google/gemini-4-hydrogenxhigh0.71125.83
- claude-fable-5-1max · not in the catalog0.71125.12
- anthropic/claude-fable-5xhigh0.71set aside
- google/gemini-3.8-flashhigh0.70114.21
- openai/gpt-5.6-solmax0.70123.35
- google/gemini-3.7-flashhigh0.69116.01
- google/gemini-3-flash-previewhigh0.67113.70
- google/gemini-3.5-flashhigh0.66111.12
- google/gemini-3.6-flashhigh0.66110.95
- openai/gpt-5.5xhigh0.63120.29
- kimi-k3.1-pro-togethermax · not in the catalog0.62118.88
- openai/gpt-6-solmax0.61124.00
- meta/muse-spark-1.2xhigh0.60117.04
- anthropic/claude-opus-5max0.60122.24
- meta/muse-spark-1.1unstated0.58set aside
- openai/gpt-6.1-terramax0.57130.85
- muse-spark-open-816b-a42bxhigh · not in the catalog0.57113.92
- x-ai/grok-4.7xhigh0.56113.32
- qwen/qwen3.7-maxunstated0.56111.06
- claude-opus-4-8max · not in the catalog0.53116.89
- deepseek/deepseek-v4-pro-0813max0.53113.60
- qwen3.6-max-previewunstated · not in the catalog0.52106.75
- claude-opus-4-7xhigh · not in the catalog0.52113.14
- moonshotai/kimi-k3max0.51114.75
- gpt-5-2025-08-07high · not in the catalog0.50108.61
- o3-2025-04-16high · not in the catalog0.49102.96
- x-ai/grok-4.6high0.49114.02
- x-ai/grok-4.6xhigh0.49114.39
- qwen3-max-2025-09-23unstated · not in the catalog0.49100.03
- x-ai/grok-4.5high0.48111.55
- gpt-5.1-2025-11-13high · not in the catalog0.48109.94
- qwen/qwen3.8-max-0902xhigh0.47113.51
- claude-opus-4-6max · not in the catalog0.47111.69
- deepseek/deepseek-v4-promax0.47107.02
- claude-sonnet-5-5max · not in the catalog0.47122.13
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.46117.92
- qwen3.8-maxxhigh · not in the catalog0.46116.39
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.46104.05
- gpt-5.4-2026-03-05xhigh · not in the catalog0.45116.23
- doubao-seed-2.1-pro-260628high · not in the catalog0.45109.61
- qwen/qwen3.6-plusunstated0.44106.49
- openai/gpt-5.6-terramax0.43119.24
- deepseek/deepseek-v4.1-flashmax0.43116.49
- deepseek/deepseek-v4.1-flash-1017max0.42116.25
- doubao-seed-evolvinghigh · not in the catalog0.42114.76
- openai/gpt-6-lunamax0.41113.99
- o1-2024-12-17high · not in the catalog0.4198.27
- openai/gpt-5.6-lunamax0.41114.95
- z-ai/glm-5.3max0.41112.54
- qwen/qwen3-235b-a22b-thinking-2507unstated0.40104.10
- thinkingmachines/inklingxhigh0.40107.98
- doubao-seed-2.0-pro-260215high · not in the catalog0.39108.12
- gpt-5.2-2025-12-11xhigh · not in the catalog0.37113.39
- moonshotai/kimi-k2.7-codeunstated0.36107.73
- claude-sonnet-4-6high · not in the catalog0.35102.30
- moonshotai/kimi-k2.6unstated0.35110.82
- moonshotai/kimi-k2.5unstated0.34set aside
- gpt-5.2-2025-12-11high · not in the catalog0.34110.74
- z-ai/glm-5.2max0.34110.33
- z-ai/glm-5.1unstated0.34109.27
- anthropic/claude-sonnet-5max0.34112.43
- deepseek/deepseek-v4-flash-0731max0.34111.14
- x-ai/grok-4.3high0.33107.39
- anthropic/claude-sonnet-5xhigh0.33113.50
- claude-sonnet-4-6max · not in the catalog0.33101.17
- gpt-5.2-2025-12-11low · not in the catalog0.33104.89
- gpt-5.2-2025-12-11medium · not in the catalog0.33108.75
- z-ai/glm-4.7unstated0.32104.07
- doubao-seed-2.0-lite-260428high · not in the catalog0.32106.42
- gpt-4.1-2025-04-14unstated · not in the catalog0.3191.64
- claude-sonnet-4-5-20250929-59kunstated · not in the catalog0.31101.26
- grok-4.20-0309-reasoningunstated · not in the catalog0.30108.33
- doubao-seed-2.1-turbo-260628high · not in the catalog0.30108.84
- gpt-5.4-mini-2026-03-17high · not in the catalog0.29106.09
- openai/gpt-4o-2024-08-06unstated0.2682.76
- qwen3.5-plusunstated · not in the catalog0.25105.31
- claude-haiku-5-5max · not in the catalog0.24113.88
- claude-sonnet-4-5-20250929unstated · not in the catalog0.2493.21
- mistral-large-4-0-osshigh · not in the catalog0.22110.78
- gpt-5-mini-2025-08-07high · not in the catalog0.22104.57
- qwen3.5-flashunstated · not in the catalog0.2099.58
- mistralai/mistral-large-4-0high0.20110.63
- o4-mini-2025-04-16low · not in the catalog0.20100.61
- thinkingmachines/inkling-smallxhigh0.19106.54
- o4-mini-2025-04-16high · not in the catalog0.19103.35
- seed-1.6unstated · not in the catalog0.1797.70
- qwen/qwen3.6-flashunstated0.16100.95
- o3-mini-2025-01-31high · not in the catalog0.15102.45
- claude-haiku-4-5-20251001unstated · not in the catalog0.1390.99
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.1393.48
- claude-3-opus-20240229unstated · not in the catalog0.1378.88
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.1396.01
- gpt-5-nano-2025-08-07high · not in the catalog0.12100.45
- gpt-5.4-nano-2026-03-17high · not in the catalog0.12106.76
- google/gemma-4-31b-itunstated0.10set aside
- openai/gpt-4o-mini-2024-07-18unstated0.0880.10
- gpt-4.1-nano-2025-04-14unstated · not in the catalog0.0686.22
FrontierMath-2025-02-28-PrivateEpoch AI Benchmarking Hub+1.00 · 2.05 · 101 runs
101 runs scored on it, 95 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-5.5-pro-pre-releasehigh · not in the catalog0.52set aside
- gpt-5.5-pre-releasexhigh · not in the catalog0.52116.86
- gpt-5.5-pro-pre-releasexhigh · not in the catalog0.51117.38
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.50117.92
- gpt-5.4-2026-03-05xhigh · not in the catalog0.48116.23
- claude-opus-4-8max · not in the catalog0.47116.89
- claude-opus-4-7xhigh · not in the catalog0.44113.14
- claude-opus-4-6max · not in the catalog0.41111.69
- gpt-5.2-2025-12-11xhigh · not in the catalog0.41113.39
- gpt-5.2-2025-12-11high · not in the catalog0.40110.74
- claude-opus-4-6-32kunstated · not in the catalog0.40111.03
- claude-opus-4-6-64kunstated · not in the catalog0.40111.92
- muse-sparkunstated · not in the catalog0.39110.41
- google/gemini-3.5-flashhigh0.39111.12
- moonshotai/kimi-k2.6unstated0.39110.82
- claude-opus-4-6unstated · not in the catalog0.38110.08
- gemini-3-pro-previewunstated · not in the catalog0.38110.74
- google/gemini-3.1-pro-previewunstated0.37111.66
- gpt-5.2-2025-12-11medium · not in the catalog0.37108.75
- google/gemini-3-flash-previewunstated0.36108.16
- z-ai/glm-5.1unstated0.33109.27
- gpt-5-2025-08-07high · not in the catalog0.32108.61
- claude-sonnet-4-6-16kunstated · not in the catalog0.32109.98
- gpt-5.1-2025-11-13high · not in the catalog0.31109.94
- gemini-2.5-deep-think-2025-08-01-webappunstated · not in the catalog0.29set aside
- gpt-5.4-mini-2026-03-17high · not in the catalog0.28106.09
- fireworks/kimi-k2p5unstated · not in the catalog0.28107.32
- gpt-5-2025-08-07medium · not in the catalog0.27108.12
- gpt-5-mini-2025-08-07high · not in the catalog0.27104.57
- gpt-5.1-2025-11-13medium · not in the catalog0.27105.41
- gpt-5.2-2025-12-11low · not in the catalog0.27104.89
- qwen/qwen3.6-plusunstated0.26106.49
- gpt-5.4-nano-2026-03-17high · not in the catalog0.26106.76
- o4-mini-2025-04-16high · not in the catalog0.25103.35
- qwen3.6-max-previewunstated · not in the catalog0.23106.75
- fireworks/deepseek-v3p2unstated · not in the catalog0.22set aside
- moonshotai/kimi-k2-thinkingunstated · not in the catalog0.21104.82
- qwen3.5-plusunstated · not in the catalog0.21105.31
- claude-opus-4-5-20251101unstated · not in the catalog0.2199.92
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.21104.05
- claude-opus-4-5-20251101-16kunstated · not in the catalog0.20104.70
- gpt-5-mini-2025-08-07medium · not in the catalog0.20103.83
- grok-4-0709unstated · not in the catalog0.20104.16
- o4-mini-2025-04-16medium · not in the catalog0.19101.93
- o3-2025-04-16high · not in the catalog0.19102.96
- gpt-5.1-2025-11-13low · not in the catalog0.17100.64
- o3-2025-04-16medium · not in the catalog0.17101.11
- z-ai/glm-5unstated0.1699.96
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.15101.18
- google/gemini-2.5-prounstated0.14103.73
- claude-sonnet-4-5-20250929-59kunstated · not in the catalog0.13101.26
- o3-mini-2025-01-31high · not in the catalog0.12102.45
- o4-mini-2025-04-16low · not in the catalog0.11100.61
- qwen/qwen3.6-flashunstated0.10100.95
- gemini-2.5-pro-preview-06-05unstated · not in the catalog0.10103.27
- o3-2025-04-16low · not in the catalog0.1099.37
- claude-sonnet-4-5-20250929unstated · not in the catalog0.0993.21
- o1-2024-12-17high · not in the catalog0.0998.27
- qwen/qwen3-235b-a22b-thinking-2507unstated · not in the catalog0.08101.29
- gpt-5-nano-2025-08-07high · not in the catalog0.08100.45
- o3-mini-2025-01-31medium · not in the catalog0.0897.94
- claude-opus-4-1-20250805-27kunstated · not in the catalog0.0797.29
- gpt-5-nano-2025-08-07medium · not in the catalog0.0799.03
- qwen3.5-flashunstated · not in the catalog0.0699.58
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.0696.01
- claude-opus-4-1-20250805unstated · not in the catalog0.0693.81
- grok-3-mini-betahigh · not in the catalog0.0697.03
- gpt-4.1-2025-04-14unstated · not in the catalog0.0691.64
- google/gemini-2.5-flashunstated0.05set aside
- claude-opus-4-20250514unstated · not in the catalog0.0492.48
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.0493.48
- claude-3-7-sonnet-20250219-16kunstated · not in the catalog0.0493.20
- claude-haiku-4-5-20251001unstated · not in the catalog0.0490.99
- claude-opus-4-20250514-27kunstated · not in the catalog0.0497.69
- claude-sonnet-4-20250514unstated · not in the catalog0.0490.16
- zai-org/glm-4.6unstated · not in the catalog0.0492.21
- grok-3-betaunstated · not in the catalog0.0494.65
- claude-3-7-sonnet-20250219-32kunstated · not in the catalog0.0394.40
- claude-3-7-sonnet-20250219unstated · not in the catalog0.0387.22
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog0.0395.18
- grok-3-mini-betalow · not in the catalog0.0394.71
- zai-org/glm-4.7unstated · not in the catalog0.02set aside
- gpt-5.1-2025-11-13none · not in the catalog0.0291.90
- claude-3-5-sonnet-20241022unstated · not in the catalog0.0285.31
- deepseek-v3unstated · not in the catalog0.0284.96
- gemini-2.0-flash-001unstated · not in the catalog0.0289.75
- o1-mini-2024-09-12medium · not in the catalog0.0291.51
- qwen-plus-2025-04-28unstated · not in the catalog0.02set aside
- o1-mini-2024-09-12high · not in the catalog0.0192.53
- claude-3-5-sonnet-20240620unstated · not in the catalog0.0181.59
- gpt-4.1-nano-2025-04-14unstated · not in the catalog0.0186.22
- qwen-max-2025-01-25unstated · not in the catalog0.0185.11
- grok-2-1212unstated · not in the catalog0.0183.89
- llama-4-maverick-17b-128e-instruct-fp8unstated · not in the catalog0.0187.21
- mistral-medium-2505unstated · not in the catalog0.0089.41
- openai/gpt-4o-2024-08-06unstated0.0082.76
- openai/gpt-4o-2024-11-20unstated0.0080.37
- claude-3-5-haiku-20241022unstated · not in the catalog0.0080.42
- mistral-large-2411unstated · not in the catalog0.0081.24
- gemini-1.5-flash-002unstated · not in the catalog083.86
- llama-4-scout-17b-16e-instructunstated · not in the catalog082.99
pythonLiveBench+0.60 · 0.52 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- deepseek/deepseek-v4.1-flashmax90116.49
- anthropic/claude-opus-5max75122.24
- deepseek/deepseek-v4-flash-vision-expunstated75116.36
- qwen/qwen3.8-27bunstated75112.72
- claude-fable-5-1max · not in the catalog70125.12
- claude-opus-5-5max · not in the catalog70128.09
- claude-opus-5-5xhigh · not in the catalog70125.14
- qwen3.8-flash-nextunstated · not in the catalog70113.12
- smaug-miniunstated · not in the catalog70114.73
- anthropic/claude-fable-5max65122.58
- moonshotai/kimi-k3unstated65116.54
- openai/gpt-6-astramax65132.41
- smaug-flashunstated · not in the catalog65116.63
- anthropic/claude-sonnet-5xhigh60113.50
- google/gemini-3.7-flashhigh60116.01
- openai/gpt-5.2-codexunstated60110.02
- openai/gpt-6.1-solmax60130.00
- openai/gpt-6.1-solxhigh60124.64
- claude-haiku-5-5max · not in the catalog60113.88
- mistral-large-4high · not in the catalog60109.81
- muse-wonder-1.0max · not in the catalog60120.86
- ox-alphamax · not in the catalog60106.05
- qwen3.8-maxunstated · not in the catalog60118.08
- smaug-agenticunstated · not in the catalog60116.30
- x-ai/grok-4.5unstated60114.74
- z-ai/glm-5.3unstated60112.92
- z-ai/glm-5.3-flashunstated60108.42
- meta/muse-spark-1.1xhigh55114.21
- meta/muse-spark-1.3xhigh55116.79
- openai/gpt-5.4-minixhigh55104.22
- openai/gpt-5.5xhigh55120.29
- openai/gpt-5.6-solmax55123.35
- openai/gpt-6-solmax55124.00
- claude-haiku-5-5xhigh · not in the catalog55112.88
- gpt-5.2-2025-12-11high · not in the catalog55110.74
- kimi-k2.6-thinkingunstated · not in the catalog55107.26
- x-ai/grok-4.6unstated55117.62
- x-ai/grok-4.7xhigh55113.32
- z-ai/glm-5.2unstated55109.40
- deepseek/deepseek-v4-flash-0731unstated50113.47
- deepseek/deepseek-v4-prounstated50111.64
- deepseek/deepseek-v4-pro-0813unstated50118.60
- google/gemini-3.1-pro-previewhigh50113.89
- google/gemini-3.5-flashhigh50111.12
- google/gemini-3.5-flash-litehigh5099.97
- google/gemini-3.8-flashhigh50114.21
- meta/muse-spark-1.2xhigh50117.04
- moonshotai/kimi-k2.7-codeunstated50107.73
- openai/gpt-5.4xhigh50119.02
- openai/gpt-5.4-nanoxhigh50110.30
- openai/gpt-5.6-terramax50119.24
- openai/gpt-6-lunamax50113.99
- thinkingmachines/inklingxhigh50107.98
- claude-opus-4-6-thinking-autohigh · not in the catalog50113.04
- claude-opus-4-8max · not in the catalog50116.89
- claude-sonnet-4-6-thinking-automedium · not in the catalog50114.06
- nvidia/nemotron-3-ultra-550b-a55bunstated45105.72
- openai/gpt-5.6-lunamax45114.95
- qwen/qwen3.6-plusunstated45106.49
- qwen/qwen3.7-maxunstated45111.06
- claude-opus-4-7xhigh · not in the catalog45113.14
- x-ai/grok-build-0.1unstated45106.25
- google/gemini-3.6-flashhigh40110.95
- qwen/qwen3.6-27bunstated40103.72
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog40110.65
- deepseek/deepseek-v4-flashunstated35102.96
- minimax/minimax-m3unstated35104.58
- claude-sonnet-5-5max · not in the catalog35122.13
- claude-sonnet-5-5xhigh · not in the catalog35119.88
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog3587.31
- x-ai/grok-4.3unstated15103.07
FrontierMath-Tiers-1-3-v2-PrivateEpoch AI Benchmarking Hub+0.58 · 2.02 · 128 runs
128 runs scored on it, 121 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-fable-5-2max · not in the catalog0.95132.06
- openai/gpt-6-astramax0.94132.41
- openai/gpt-6.1-solmax0.94130.00
- gpt-6.2-astramax · not in the catalog0.93135.84
- claude-opus-5-5max · not in the catalog0.91128.09
- gemini-4-neonxhigh · not in the catalog0.91134.54
- openai/gpt-6.1-terramax0.91130.85
- google/gemini-4-carbonhigh0.91131.68
- claude-fable-5-1max · not in the catalog0.90125.12
- openai/gpt-6-solmax0.90124.00
- openai/gpt-5.6-solmax0.89123.35
- claude-sonnet-5-5max · not in the catalog0.89122.13
- google/gemini-4-hydrogenxhigh0.88125.83
- openai/gpt-5.5-proxhigh0.88122.66
- anthropic/claude-fable-5max0.87122.58
- google/gemini-4-argonhigh0.86124.90
- openai/gpt-5.6-terramax0.86119.24
- anthropic/claude-opus-5max0.86122.24
- openai/gpt-5.5xhigh0.85120.29
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.82117.92
- openai/gpt-5.6-lunamax0.82114.95
- muse-wonder-1.0max · not in the catalog0.82120.86
- claude-opus-4-8max · not in the catalog0.80116.89
- openai/gpt-6-lunamax0.79113.99
- gpt-5.4-2026-03-05xhigh · not in the catalog0.79116.23
- kimi-k3.1-pro-togethermax · not in the catalog0.78118.88
- claude-haiku-5-5max · not in the catalog0.75113.88
- qwen3.8-maxxhigh · not in the catalog0.75116.39
- meta/muse-spark-1.3xhigh0.74116.79
- meta/muse-spark-1.3max0.74115.55
- gpt-5.2-pro-2025-12-11xhigh · not in the catalog0.74116.74
- moonshotai/kimi-k3max0.72114.75
- google/gemini-3.7-flashhigh0.72116.01
- deepseek/deepseek-v4.1-flash-1017max0.71116.25
- claude-opus-4-7max · not in the catalog0.70113.03
- muse-spark-open-816b-a42bxhigh · not in the catalog0.70113.92
- z-ai/glm-5.3max0.69112.54
- google/gemini-3.8-flashhigh0.68114.21
- gpt-5.2-2025-12-11xhigh · not in the catalog0.67113.39
- deepseek/deepseek-v4.1-flashmax0.67116.49
- claude-opus-4-6max · not in the catalog0.66111.69
- x-ai/grok-4.6xhigh0.66114.39
- anthropic/claude-sonnet-5max0.66112.43
- qwen/qwen3.8-max-0902xhigh0.66113.51
- deepseek/deepseek-v4-pro-0813max0.65113.60
- qwen/qwen3.7-maxunstated0.65111.06
- google/gemini-3.5-flashhigh0.63111.12
- google/gemini-3.1-pro-previewunstated0.60111.66
- z-ai/glm-5.2max0.59110.33
- google/gemini-3.6-flashhigh0.59110.95
- mistralai/mistral-large-4-0high0.59110.63
- deepseek/deepseek-v4-flash-0731max0.58111.14
- moonshotai/kimi-k2.6unstated0.57110.82
- x-ai/grok-4.5high0.57111.55
- gpt-5-pro-2025-10-06high · not in the catalog0.56109.23
- z-ai/glm-5.3-flashmax0.56108.45
- gpt-5-2025-08-07high · not in the catalog0.55108.61
- z-ai/glm-5.2low0.55105.42
- moonshotai/kimi-k2.7-codeunstated0.54107.73
- x-ai/grok-4.7xhigh0.53113.32
- google/gemini-3-flash-previewunstated0.51108.16
- gpt-5.4-mini-2026-03-17xhigh · not in the catalog0.51106.91
- gpt-5-mini-2025-08-07high · not in the catalog0.47104.57
- thinkingmachines/inkling-smallxhigh0.46106.54
- deepseek/deepseek-v4-promax0.45107.02
- gpt-5.4-nano-2026-03-17high · not in the catalog0.45106.76
- grok-4.20-0309-reasoningunstated · not in the catalog0.45108.33
- x-ai/grok-4.3high0.43107.39
- z-ai/glm-5.2none0.4296.22
- openai/gpt-5.6-lunalow0.41101.92
- openai/gpt-5.6-lunanone0.4095.93
- qwen/qwen3.6-plusunstated0.38106.49
- gpt-5-2025-08-07low · not in the catalog0.37105.22
- z-ai/glm-5.1unstated0.37109.27
- o4-mini-2025-04-16high · not in the catalog0.36103.35
- qwen/qwen3.6-27bunstated0.35103.72
- qwen/qwen3.7-plusnone0.34103.17
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.34104.05
- qwen/qwen3.6-27bnone0.34100.32
- thinkingmachines/inklingxhigh0.33107.98
- o3-2025-04-16high · not in the catalog0.33102.96
- qwen/qwen3.6-plusnone0.32set aside
- qwen/qwen3.5-397b-a17bnone0.31103.55
- o3-2025-04-16medium · not in the catalog0.30101.11
- qwen/qwen3.5-397b-a17bunstated0.29104.40
- o4-mini-2025-04-16medium · not in the catalog0.29101.93
- google/gemini-3.1-flash-litehigh0.28102.13
- gpt-5.5-instantunstated · not in the catalog0.2699.89
- google/gemini-3.5-flash-litehigh0.2699.97
- z-ai/glm-5.1none0.25set aside
- google/gemini-2.5-prounstated0.25103.73
- gpt-5.4-mini-2026-03-17low · not in the catalog0.25set aside
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.24101.18
- google/gemini-3.1-flash-litelow0.2294.94
- qwen/qwen3.6-flashnone0.22set aside
- google/gemini-3.1-flash-liteminimal0.2193.68
- qwen/qwen3.6-35b-a3bnone0.20100.22
- gpt-5.4-nano-2026-03-17low · not in the catalog0.2098.40
- gpt-5-nano-2025-08-07high · not in the catalog0.20100.45
- qwen/qwen3.7-flashunstated0.19102.16
- qwen/qwen3.7-flashnone0.19100.54
- o3-2025-04-16low · not in the catalog0.1999.37
- qwen3-max-2025-09-23unstated · not in the catalog0.19100.03
- o3-mini-2025-01-31high · not in the catalog0.19102.45
- gpt-5-2025-08-07minimal · not in the catalog0.1895.14
- gpt-5-mini-2025-08-07low · not in the catalog0.18set aside
- qwen3.5-flashnone · not in the catalog0.18set aside
- qwen/qwen3.6-35b-a3bunstated0.18102.04
- qwen/qwen3.6-flashunstated0.17100.95
- gpt-5.4-mini-2026-03-17none · not in the catalog0.1790.48
- o4-mini-2025-04-16low · not in the catalog0.16100.61
- o1-2024-12-17high · not in the catalog0.1598.27
- claude-opus-4-1-20250805-32kunstated · not in the catalog0.13set aside
- o3-mini-2025-01-31medium · not in the catalog0.1197.94
- o1-2024-12-17medium · not in the catalog0.1097.81
- qwen3.5-flashunstated · not in the catalog0.0999.58
- o1-2024-12-17low · not in the catalog0.0894.71
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.0793.48
- gpt-4.1-2025-04-14unstated · not in the catalog0.0691.64
- gpt-5-mini-2025-08-07minimal · not in the catalog0.0693.86
- gpt-5-nano-2025-08-07low · not in the catalog0.0692.74
- gpt-5.4-nano-2026-03-17none · not in the catalog0.0591.33
- o3-mini-2025-01-31low · not in the catalog0.0491.91
- gpt-5-nano-2025-08-07minimal · not in the catalog0.0288.08
- openai/gpt-4o-mini-2024-07-18unstated0.0180.10
- gpt-4-turbo-2024-04-09unstated · not in the catalog0.0180.07
- openai/gpt-4o-2024-08-06unstated0.0082.76
- gpt-3.5-turbo-0125unstated · not in the catalog069.92
FrontierMath-2025-02-28-PublicEpoch AI Benchmarking Hub+0.43 · 4.76 · 87 runs
87 runs scored on it, 85 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5max1122.58
- moonshotai/kimi-k2.6unstated0.90110.82
- claude-opus-4-6max · not in the catalog0.90111.69
- claude-opus-4-6-64kunstated · not in the catalog0.90111.92
- claude-opus-4-7xhigh · not in the catalog0.90113.14
- google/gemini-3.1-pro-previewunstated0.89111.66
- google/gemini-3.5-flashhigh0.80111.12
- claude-opus-4-6-32kunstated · not in the catalog0.80111.03
- claude-sonnet-4-6-16kunstated · not in the catalog0.80109.98
- gemini-3-pro-previewunstated · not in the catalog0.80110.74
- gpt-5.4-2026-03-05high · not in the catalog0.80115.98
- z-ai/glm-5.1unstated0.80109.27
- gpt-5-2025-08-07medium · not in the catalog0.70108.12
- google/gemini-3-flash-previewunstated0.60108.16
- gpt-5-2025-08-07high · not in the catalog0.60108.61
- gpt-5-pro-2025-10-06high · not in the catalog0.60109.23
- gpt-5.2-2025-12-11high · not in the catalog0.60110.74
- gpt-5.2-2025-12-11medium · not in the catalog0.60108.75
- gpt-5.4-nano-2026-03-17high · not in the catalog0.60106.76
- qwen/qwen3.6-plusunstated0.50106.49
- gpt-5.4-mini-2026-03-17high · not in the catalog0.50106.09
- qwen3.5-plusunstated · not in the catalog0.50105.31
- qwen3.6-max-previewunstated · not in the catalog0.50106.75
- google/gemini-2.5-prounstated0.40103.73
- claude-opus-4-5-20251101-16kunstated · not in the catalog0.40104.70
- gpt-5-mini-2025-08-07medium · not in the catalog0.40103.83
- gpt-5.2-2025-12-11low · not in the catalog0.40104.89
- moonshotai/kimi-k2-thinkingunstated · not in the catalog0.40104.82
- o3-mini-2025-01-31high · not in the catalog0.40102.45
- gemini-2.5-pro-preview-06-05unstated · not in the catalog0.30103.27
- gpt-5-mini-2025-08-07high · not in the catalog0.30104.57
- o4-mini-2025-04-16high · not in the catalog0.30103.35
- o4-mini-2025-04-16low · not in the catalog0.30100.61
- qwen/qwen3.6-flashunstated0.20100.95
- claude-opus-4-5-20251101unstated · not in the catalog0.2099.92
- gpt-5-nano-2025-08-07high · not in the catalog0.20100.45
- kimi-k2-thinking-turbounstated · not in the catalog0.20102.46
- o4-mini-2025-04-16medium · not in the catalog0.20101.93
- qwen/qwen3-235b-a22b-thinking-2507unstated · not in the catalog0.20101.29
- o3-mini-2025-01-31medium · not in the catalog0.1197.94
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.1093.48
- gpt-5-nano-2025-08-07medium · not in the catalog0.1099.03
- o3-2025-04-16high · not in the catalog0.10102.96
- o3-2025-04-16low · not in the catalog0.1099.37
- qwen3.5-flashunstated · not in the catalog0.1099.58
- z-ai/glm-5unstated0.0299.96
- openai/gpt-4o-2024-08-06unstated082.76
- openai/gpt-4o-2024-11-20unstated080.37
- claude-3-5-haiku-20241022unstated · not in the catalog080.42
- claude-3-5-sonnet-20240620unstated · not in the catalog081.59
- claude-3-5-sonnet-20241022unstated · not in the catalog085.31
- claude-3-7-sonnet-20250219unstated · not in the catalog087.22
- claude-3-7-sonnet-20250219-16kunstated · not in the catalog093.20
- claude-3-7-sonnet-20250219-32kunstated · not in the catalog094.40
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog095.18
- claude-haiku-4-5-20251001-32kunstated · not in the catalog096.01
- claude-opus-4-1-20250805unstated · not in the catalog093.81
- claude-opus-4-1-20250805-16kunstated · not in the catalog096.39
- claude-opus-4-1-20250805-27kunstated · not in the catalog097.29
- claude-opus-4-20250514unstated · not in the catalog092.48
- claude-sonnet-4-20250514unstated · not in the catalog090.16
- claude-sonnet-4-5-20250929unstated · not in the catalog093.21
- deepseek-v3unstated · not in the catalog084.96
- deepseek-v3-0324unstated · not in the catalog090.12
- gemini-1.5-flash-002unstated · not in the catalog083.86
- gemini-2.0-flash-001unstated · not in the catalog089.75
- gemini-2.0-pro-exp-02-05unstated · not in the catalog090.84
- gpt-4.1-2025-04-14unstated · not in the catalog091.64
- gpt-4.1-nano-2025-04-14unstated · not in the catalog086.22
- grok-2-1212unstated · not in the catalog083.89
- grok-3-betaunstated · not in the catalog094.65
- grok-3-mini-betahigh · not in the catalog097.03
- grok-3-mini-betalow · not in the catalog094.71
- llama-4-maverick-17b-128e-instruct-fp8unstated · not in the catalog087.21
- llama-4-scout-17b-16e-instructunstated · not in the catalog082.99
- magistral-small-2506unstated · not in the catalog088.63
- mistral-large-2411unstated · not in the catalog081.24
- mistral-medium-2505unstated · not in the catalog089.41
- o1-2024-12-17high · not in the catalog098.27
- o1-mini-2024-09-12high · not in the catalog092.53
- o1-mini-2024-09-12medium · not in the catalog091.51
- o3-2025-04-16medium · not in the catalog0101.11
- qwen-max-2025-01-25unstated · not in the catalog085.11
- qwen-plus-2025-04-28unstated · not in the catalog0set aside
- qwen3-235b-a22bunstated · not in the catalog087.56
- zai-org/glm-4.6unstated · not in the catalog092.21
- z-ai/glm-4.5unstated0set aside
plot_unscramblingLiveBench+0.42 · 0.89 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-6-astramax86.29132.41
- openai/gpt-6.1-solmax84.38130.00
- openai/gpt-6.1-solxhigh81.94124.64
- openai/gpt-5.6-solmax79.05123.35
- anthropic/claude-fable-5max78.72122.58
- openai/gpt-6-solmax77.90124.00
- claude-opus-5-5max · not in the catalog77.48128.09
- claude-opus-5-5xhigh · not in the catalog77.16125.14
- claude-fable-5-1max · not in the catalog75.17125.12
- muse-wonder-1.0max · not in the catalog74.80120.86
- anthropic/claude-opus-5max74.73122.24
- google/gemini-3.1-pro-previewhigh74.13113.89
- openai/gpt-5.5xhigh74.09120.29
- google/gemini-3.8-flashhigh73.38114.21
- moonshotai/kimi-k3unstated72.58116.54
- x-ai/grok-4.7xhigh72.42113.32
- smaug-agenticunstated · not in the catalog71.07116.30
- openai/gpt-5.6-terramax69.68119.24
- meta/muse-spark-1.3xhigh68.38116.79
- claude-sonnet-5-5xhigh · not in the catalog67.79119.88
- google/gemini-3.5-flashhigh67.75111.12
- x-ai/grok-4.6unstated67.09117.62
- claude-opus-4-6-thinking-autohigh · not in the catalog66.48113.04
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog66.45110.65
- x-ai/grok-4.5unstated66.39114.74
- google/gemini-3.7-flashhigh66.36116.01
- openai/gpt-5.4xhigh65.90119.02
- deepseek/deepseek-v4-pro-0813unstated64.22118.60
- google/gemini-3.6-flashhigh63.69110.95
- meta/muse-spark-1.1xhigh63.03114.21
- meta/muse-spark-1.2xhigh61.72117.04
- claude-opus-4-8max · not in the catalog61.65116.89
- deepseek/deepseek-v4.1-flashmax61.57116.49
- claude-opus-4-7xhigh · not in the catalog61.41113.14
- z-ai/glm-5.2unstated60.73109.40
- smaug-flashunstated · not in the catalog60.31116.63
- z-ai/glm-5.3-flashunstated60.26108.42
- z-ai/glm-5.3unstated60.24112.92
- qwen/qwen3.7-maxunstated58.72111.06
- qwen3.8-maxunstated · not in the catalog58.56118.08
- gpt-5.2-2025-12-11high · not in the catalog58.43110.74
- deepseek/deepseek-v4-flash-0731unstated58.19113.47
- kimi-k2.6-thinkingunstated · not in the catalog58.09107.26
- deepseek/deepseek-v4-flash-vision-expunstated57.75116.36
- openai/gpt-6-lunamax57.48113.99
- claude-sonnet-4-6-thinking-automedium · not in the catalog56.98114.06
- qwen/qwen3.8-27bunstated56.54112.72
- deepseek/deepseek-v4-prounstated56.39111.64
- openai/gpt-5.2-codexunstated56.03110.02
- qwen3.8-flash-nextunstated · not in the catalog55.76113.12
- moonshotai/kimi-k2.7-codeunstated55.72107.73
- smaug-miniunstated · not in the catalog55.33114.73
- anthropic/claude-sonnet-5xhigh53.41113.50
- x-ai/grok-build-0.1unstated53.38106.25
- thinkingmachines/inklingxhigh53.21107.98
- ox-alphamax · not in the catalog53.08106.05
- qwen/qwen3.6-plusunstated52.47106.49
- openai/gpt-5.6-lunamax51.20114.95
- minimax/minimax-m3unstated50.17104.58
- google/gemini-3.5-flash-litehigh49.4699.97
- x-ai/grok-4.3unstated48.74103.07
- claude-sonnet-5-5max · not in the catalog48.62122.13
- nvidia/nemotron-3-ultra-550b-a55bunstated47.77105.72
- deepseek/deepseek-v4-flashunstated46.87102.96
- openai/gpt-5.4-minixhigh43.88104.22
- qwen/qwen3.6-27bunstated42.75103.72
- claude-haiku-5-5xhigh · not in the catalog42.72112.88
- mistral-large-4high · not in the catalog40.93109.81
- openai/gpt-5.4-nanoxhigh39.13110.30
- claude-haiku-5-5max · not in the catalog27.88113.88
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog22.8787.31
integrals_with_gameLiveBench+0.28 · 2.15 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-5.6-solmax100123.35
- openai/gpt-6-astramax100132.41
- openai/gpt-6.1-solmax100130.00
- claude-opus-5-5max · not in the catalog100128.09
- claude-sonnet-5-5xhigh · not in the catalog100119.88
- openai/gpt-5.5xhigh99120.29
- openai/gpt-6-solmax99124.00
- openai/gpt-6.1-solxhigh99124.64
- claude-fable-5-1max · not in the catalog99125.12
- claude-opus-5-5xhigh · not in the catalog99125.14
- muse-wonder-1.0max · not in the catalog98120.86
- anthropic/claude-fable-5max97122.58
- anthropic/claude-opus-5max97122.24
- claude-sonnet-5-5max · not in the catalog97122.13
- meta/muse-spark-1.3xhigh96116.79
- x-ai/grok-4.7xhigh96113.32
- openai/gpt-5.6-terramax95119.24
- deepseek/deepseek-v4-pro-0813unstated94118.60
- openai/gpt-5.4xhigh93119.02
- gpt-5.2-2025-12-11high · not in the catalog92110.74
- meta/muse-spark-1.2xhigh91117.04
- deepseek/deepseek-v4.1-flashmax90116.49
- claude-sonnet-4-6-thinking-automedium · not in the catalog90114.06
- mistral-large-4high · not in the catalog90109.81
- claude-haiku-5-5xhigh · not in the catalog89112.88
- claude-opus-4-8max · not in the catalog89116.89
- anthropic/claude-sonnet-5xhigh88113.50
- google/gemini-3.7-flashhigh88116.01
- x-ai/grok-4.6unstated85117.62
- claude-opus-4-7xhigh · not in the catalog84113.14
- meta/muse-spark-1.1xhigh83114.21
- openai/gpt-5.4-nanoxhigh82.83110.30
- qwen3.8-maxunstated · not in the catalog81118.08
- google/gemini-3.8-flashhigh80114.21
- nvidia/nemotron-3-ultra-550b-a55bunstated80105.72
- deepseek/deepseek-v4-prounstated78111.64
- google/gemini-3.1-pro-previewhigh78113.89
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog78110.65
- x-ai/grok-4.5unstated78114.74
- deepseek/deepseek-v4-flash-vision-expunstated76116.36
- smaug-flashunstated · not in the catalog76116.63
- smaug-miniunstated · not in the catalog76114.73
- z-ai/glm-5.2unstated76109.40
- openai/gpt-5.2-codexunstated75110.02
- openai/gpt-6-lunamax74113.99
- claude-opus-4-6-thinking-autohigh · not in the catalog73113.04
- thinkingmachines/inklingxhigh72107.98
- openai/gpt-5.6-lunamax70114.95
- z-ai/glm-5.3unstated69112.92
- google/gemini-3.5-flashhigh68111.12
- claude-haiku-5-5max · not in the catalog67113.88
- x-ai/grok-build-0.1unstated66106.25
- deepseek/deepseek-v4-flash-0731unstated65113.47
- qwen/qwen3.8-27bunstated65112.72
- google/gemini-3.6-flashhigh63110.95
- x-ai/grok-4.3unstated63103.07
- qwen/qwen3.6-plusunstated62106.49
- qwen/qwen3.7-maxunstated59111.06
- qwen3.8-flash-nextunstated · not in the catalog59113.12
- minimax/minimax-m3unstated56104.58
- moonshotai/kimi-k3unstated54116.54
- kimi-k2.6-thinkingunstated · not in the catalog54107.26
- qwen/qwen3.6-27bunstated52103.72
- smaug-agenticunstated · not in the catalog52116.30
- z-ai/glm-5.3-flashunstated45108.42
- ox-alphamax · not in the catalog39106.05
- deepseek/deepseek-v4-flashunstated37102.96
- moonshotai/kimi-k2.7-codeunstated37107.73
- openai/gpt-5.4-minixhigh37104.22
- google/gemini-3.5-flash-litehigh2699.97
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog2487.31
consecutive_eventsLiveBench+0.16 · 1.44 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- muse-wonder-1.0max · not in the catalog98.53120.86
- claude-sonnet-4-6-thinking-automedium · not in the catalog92.68114.06
- anthropic/claude-fable-5max91.36122.58
- openai/gpt-6.1-solxhigh91.29124.64
- claude-sonnet-5-5xhigh · not in the catalog90.94119.88
- openai/gpt-6-astramax90.56132.41
- openai/gpt-6.1-solmax90.56130.00
- claude-fable-5-1max · not in the catalog90.56125.12
- claude-opus-5-5max · not in the catalog90.56128.09
- meta/muse-spark-1.3xhigh90.50116.79
- smaug-agenticunstated · not in the catalog90.42116.30
- claude-opus-5-5xhigh · not in the catalog90.37125.14
- deepseek/deepseek-v4-pro-0813unstated90.23118.60
- openai/gpt-5.6-solmax90.21123.35
- deepseek/deepseek-v4-flash-vision-expunstated90.05116.36
- moonshotai/kimi-k3unstated89.76116.54
- openai/gpt-6-solmax89.55124.00
- claude-opus-4-7xhigh · not in the catalog89.52113.14
- deepseek/deepseek-v4-flash-0731unstated89.44113.47
- openai/gpt-5.2-codexunstated89.02110.02
- gpt-5.2-2025-12-11high · not in the catalog88.91110.74
- openai/gpt-5.5xhigh88.82120.29
- deepseek/deepseek-v4.1-flashmax88.72116.49
- smaug-flashunstated · not in the catalog88.61116.63
- smaug-miniunstated · not in the catalog88.50114.73
- openai/gpt-5.6-terramax87.67119.24
- qwen3.8-maxunstated · not in the catalog87.14118.08
- openai/gpt-5.6-lunamax86.50114.95
- openai/gpt-5.4xhigh86.19119.02
- x-ai/grok-4.7xhigh85.73113.32
- google/gemini-3.1-pro-previewhigh85.24113.89
- qwen/qwen3.8-27bunstated84.93112.72
- minimax/minimax-m3unstated84.61104.58
- z-ai/glm-5.3-flashunstated84.30108.42
- meta/muse-spark-1.2xhigh82.55117.04
- mistral-large-4high · not in the catalog80.53109.81
- ox-alphamax · not in the catalog80.35106.05
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog79.36110.65
- z-ai/glm-5.2unstated79.28109.40
- x-ai/grok-4.6unstated78.15117.62
- anthropic/claude-opus-5max77.57122.24
- qwen3.8-flash-nextunstated · not in the catalog75.62113.12
- x-ai/grok-4.5unstated75.54114.74
- deepseek/deepseek-v4-prounstated75.42111.64
- anthropic/claude-sonnet-5xhigh74.74113.50
- z-ai/glm-5.3unstated72.31112.92
- qwen/qwen3.7-maxunstated71.80111.06
- meta/muse-spark-1.1xhigh71.09114.21
- claude-haiku-5-5xhigh · not in the catalog71.05112.88
- openai/gpt-6-lunamax70.49113.99
- qwen/qwen3.6-27bunstated70.39103.72
- thinkingmachines/inklingxhigh69.85107.98
- x-ai/grok-build-0.1unstated69.73106.25
- qwen/qwen3.6-plusunstated67.04106.49
- claude-opus-4-6-thinking-autohigh · not in the catalog63.47113.04
- openai/gpt-5.4-minixhigh62.97104.22
- google/gemini-3.7-flashhigh60.37116.01
- deepseek/deepseek-v4-flashunstated59.91102.96
- openai/gpt-5.4-nanoxhigh54.30110.30
- claude-opus-4-8max · not in the catalog52.91116.89
- kimi-k2.6-thinkingunstated · not in the catalog51.27107.26
- moonshotai/kimi-k2.7-codeunstated48.28107.73
- google/gemini-3.5-flashhigh47.99111.12
- google/gemini-3.6-flashhigh43.82110.95
- claude-sonnet-5-5max · not in the catalog40.62122.13
- x-ai/grok-4.3unstated28.03103.07
- google/gemini-3.5-flash-litehigh21.0999.97
- google/gemini-3.8-flashhigh20.61114.21
- nvidia/nemotron-3-ultra-550b-a55bunstated18.77105.72
- claude-haiku-5-5max · not in the catalog10.64113.88
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog2.5187.31
zebra_puzzleLiveBench+0.13 · 3.39 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5max100122.58
- anthropic/claude-opus-5max100122.24
- deepseek/deepseek-v4-flash-0731unstated100113.47
- deepseek/deepseek-v4.1-flashmax100116.49
- moonshotai/kimi-k3unstated100116.54
- openai/gpt-5.5xhigh100120.29
- openai/gpt-5.6-solmax100123.35
- openai/gpt-5.6-terramax100119.24
- openai/gpt-6-astramax100132.41
- openai/gpt-6-solmax100124.00
- openai/gpt-6.1-solmax100130.00
- openai/gpt-6.1-solxhigh100124.64
- claude-fable-5-1max · not in the catalog100125.12
- claude-opus-4-8max · not in the catalog100116.89
- claude-opus-5-5max · not in the catalog100128.09
- claude-opus-5-5xhigh · not in the catalog100125.14
- claude-sonnet-5-5max · not in the catalog100122.13
- claude-sonnet-5-5xhigh · not in the catalog100119.88
- muse-wonder-1.0max · not in the catalog100120.86
- qwen3.8-maxunstated · not in the catalog100118.08
- deepseek/deepseek-v4-flash-vision-expunstated98.75116.36
- qwen3.8-flash-nextunstated · not in the catalog98.75113.12
- google/gemini-3.8-flashhigh98.25114.21
- smaug-agenticunstated · not in the catalog98.25116.30
- meta/muse-spark-1.3xhigh98116.79
- openai/gpt-5.4xhigh98119.02
- claude-haiku-5-5xhigh · not in the catalog98112.88
- claude-opus-4-7xhigh · not in the catalog98113.14
- smaug-flashunstated · not in the catalog98116.63
- openai/gpt-5.4-nanoxhigh97.62110.30
- deepseek/deepseek-v4-pro-0813unstated96.75118.60
- qwen/qwen3.8-27bunstated96.50112.72
- meta/muse-spark-1.1xhigh96114.21
- moonshotai/kimi-k2.7-codeunstated96107.73
- openai/gpt-6-lunamax96113.99
- claude-sonnet-4-6-thinking-automedium · not in the catalog96114.06
- x-ai/grok-4.6unstated95.50117.62
- meta/muse-spark-1.2xhigh95.25117.04
- google/gemini-3.7-flashhigh94.50116.01
- claude-haiku-5-5max · not in the catalog94113.88
- claude-opus-4-6-thinking-autohigh · not in the catalog94113.04
- smaug-miniunstated · not in the catalog94114.73
- x-ai/grok-4.5unstated94114.74
- openai/gpt-5.6-lunamax93.50114.95
- mistral-large-4high · not in the catalog93109.81
- deepseek/deepseek-v4-prounstated92111.64
- z-ai/glm-5.3unstated88.75112.92
- anthropic/claude-sonnet-5xhigh88113.50
- thinkingmachines/inklingxhigh88107.98
- x-ai/grok-4.7xhigh86113.32
- google/gemini-3.6-flashhigh85.75110.95
- google/gemini-3.1-pro-previewhigh85.25113.89
- gpt-5.2-2025-12-11high · not in the catalog84110.74
- ox-alphamax · not in the catalog79.50106.05
- z-ai/glm-5.3-flashunstated79.50108.42
- kimi-k2.6-thinkingunstated · not in the catalog78.50107.26
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog77.50110.65
- google/gemini-3.5-flashhigh77.25111.12
- qwen/qwen3.7-maxunstated74.50111.06
- nvidia/nemotron-3-ultra-550b-a55bunstated71.25105.72
- openai/gpt-5.2-codexunstated70110.02
- x-ai/grok-build-0.1unstated68.25106.25
- qwen/qwen3.6-plusunstated68106.49
- z-ai/glm-5.2unstated65.50109.40
- minimax/minimax-m3unstated61104.58
- openai/gpt-5.4-minixhigh60.37104.22
- x-ai/grok-4.3unstated57.75103.07
- qwen/qwen3.6-27bunstated53.75103.72
- deepseek/deepseek-v4-flashunstated49.25102.96
- google/gemini-3.5-flash-litehigh32.7599.97
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog17.2587.31
javascriptLiveBench-0.07 · 0.58 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- deepseek/deepseek-v4.1-flashmax81.82116.49
- claude-opus-5-5max · not in the catalog81.82128.09
- anthropic/claude-opus-5max77.27122.24
- meta/muse-spark-1.1xhigh77.27114.21
- meta/muse-spark-1.3xhigh77.27116.79
- claude-sonnet-5-5max · not in the catalog77.27122.13
- qwen3.8-maxunstated · not in the catalog77.27118.08
- smaug-agenticunstated · not in the catalog77.27116.30
- google/gemini-3.8-flashhigh72.73114.21
- meta/muse-spark-1.2xhigh72.73117.04
- claude-opus-5-5xhigh · not in the catalog72.73125.14
- x-ai/grok-4.5unstated72.73114.74
- x-ai/grok-4.6unstated72.73117.62
- z-ai/glm-5.3unstated72.73112.92
- anthropic/claude-fable-5max68.18122.58
- anthropic/claude-sonnet-5xhigh68.18113.50
- deepseek/deepseek-v4-pro-0813unstated68.18118.60
- google/gemini-3.7-flashhigh68.18116.01
- moonshotai/kimi-k3unstated68.18116.54
- openai/gpt-5.2-codexunstated68.18110.02
- openai/gpt-5.4xhigh68.18119.02
- openai/gpt-5.6-terramax68.18119.24
- thinkingmachines/inklingxhigh68.18107.98
- claude-fable-5-1max · not in the catalog68.18125.12
- claude-haiku-5-5max · not in the catalog68.18113.88
- claude-opus-4-8max · not in the catalog68.18116.89
- mistral-large-4high · not in the catalog68.18109.81
- muse-wonder-1.0max · not in the catalog68.18120.86
- qwen3.8-flash-nextunstated · not in the catalog68.18113.12
- smaug-flashunstated · not in the catalog68.18116.63
- deepseek/deepseek-v4-flash-0731unstated63.64113.47
- deepseek/deepseek-v4-flash-vision-expunstated63.64116.36
- google/gemini-3.5-flashhigh63.64111.12
- google/gemini-3.6-flashhigh63.64110.95
- minimax/minimax-m3unstated63.64104.58
- moonshotai/kimi-k2.7-codeunstated63.64107.73
- openai/gpt-5.4-nanoxhigh63.64110.30
- openai/gpt-5.5xhigh63.64120.29
- openai/gpt-5.6-lunamax63.64114.95
- openai/gpt-5.6-solmax63.64123.35
- openai/gpt-6-astramax63.64132.41
- openai/gpt-6-lunamax63.64113.99
- openai/gpt-6-solmax63.64124.00
- openai/gpt-6.1-solmax63.64130.00
- openai/gpt-6.1-solxhigh63.64124.64
- claude-opus-4-6-thinking-autohigh · not in the catalog63.64113.04
- claude-opus-4-7xhigh · not in the catalog63.64113.14
- x-ai/grok-4.7xhigh63.64113.32
- z-ai/glm-5.2unstated63.64109.40
- z-ai/glm-5.3-flashunstated63.64108.42
- google/gemini-3.1-pro-previewhigh59.09113.89
- google/gemini-3.5-flash-litehigh59.0999.97
- qwen/qwen3.6-plusunstated59.09106.49
- qwen/qwen3.7-maxunstated59.09111.06
- qwen/qwen3.8-27bunstated59.09112.72
- claude-haiku-5-5xhigh · not in the catalog59.09112.88
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog59.09110.65
- gpt-5.2-2025-12-11high · not in the catalog59.09110.74
- kimi-k2.6-thinkingunstated · not in the catalog59.09107.26
- smaug-miniunstated · not in the catalog59.09114.73
- x-ai/grok-build-0.1unstated59.09106.25
- deepseek/deepseek-v4-flashunstated54.55102.96
- deepseek/deepseek-v4-prounstated54.55111.64
- nvidia/nemotron-3-ultra-550b-a55bunstated54.55105.72
- qwen/qwen3.6-27bunstated54.55103.72
- claude-sonnet-4-6-thinking-automedium · not in the catalog54.55114.06
- ox-alphamax · not in the catalog54.55106.05
- openai/gpt-5.4-minixhigh50104.22
- claude-sonnet-5-5xhigh · not in the catalog36.36119.88
- x-ai/grok-4.3unstated27.27103.07
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog18.1887.31
polyglot pass rateAider polyglot-0.23 · 1.58 · 69 runs
69 runs scored on it, 17 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-5high88set aside
- openai/gpt-5medium86.70set aside
- openai/o3-prohigh84.90set aside
- gemini-2.5-pro-preview-06-05-(32k-think)unstated · not in the catalog83.10set aside
- openai/gpt-5low81.30set aside
- openai/o3high81.30set aside
- grok-4high · not in the catalog79.60set aside
- gemini-2.5-pro-preview-06-05-(default-think)unstated · not in the catalog79.10set aside
- o3-(high)-+-gpt-4.1unstated · not in the catalog78.20set aside
- openai/o3unstated76.90set aside
- gemini-2.5-pro-preview-05-06unstated · not in the catalog76.90102.43
- deepseek-v3.2-exp-(reasoner)unstated · not in the catalog74.20set aside
- gemini-2.5-pro-preview-03-25unstated · not in the catalog72.90set aside
- claude-opus-4-20250514-(32k-thinking)unstated · not in the catalog72set aside
- o4-minihigh · not in the catalog72set aside
- deepseek-r1-(0528)unstated · not in the catalog71.40set aside
- claude-opus-4-20250514-(no-think)unstated · not in the catalog70.70set aside
- deepseek-v3.2-exp-(chat)unstated · not in the catalog70.20set aside
- claude-3-7-sonnet-20250219-(32k-thinking-tokens)unstated · not in the catalog64.90set aside
- deepseek-r1-+-claude-3-5-sonnet-20241022unstated · not in the catalog64set aside
- o1-2024-12-17high · not in the catalog61.7098.27
- claude-sonnet-4-20250514-(32k-thinking)unstated · not in the catalog61.30set aside
- claude-3-7-sonnet-20250219-(no-thinking)unstated · not in the catalog60.40set aside
- o3-minihigh · not in the catalog60.40set aside
- qwen3-235b-a22b-diff,-no-think,-alibaba-apiunstated · not in the catalog59.60set aside
- moonshotai/kimi-k2unstated59.10set aside
- deepseek-r1unstated · not in the catalog56.9096.08
- claude-sonnet-4-20250514-(no-thinking)unstated · not in the catalog56.40set aside
- deepseek-v3-(0324)unstated · not in the catalog55.10set aside
- gemini-2.5-flash-preview-05-20-(24k-think)unstated · not in the catalog55.10set aside
- quasar-alphaunstated · not in the catalog54.70set aside
- o3-minimedium · not in the catalog53.80set aside
- grok-3-betaunstated · not in the catalog53.3094.65
- optimus-alphaunstated · not in the catalog52.90set aside
- openai/gpt-4.1unstated52.40set aside
- claude-3-5-sonnet-20241022unstated · not in the catalog51.6085.31
- grok-3-mini-betahigh · not in the catalog49.3097.03
- deepseek-chat-v3-(prev)unstated · not in the catalog48.40set aside
- gemini-2.5-flash-preview-04-17-(default)unstated · not in the catalog47.10set aside
- chatgpt-4o-latest-(2025-03-29)unstated · not in the catalog45.30set aside
- gpt-4.5-previewunstated · not in the catalog44.90set aside
- gemini-2.5-flash-preview-05-20-(no-think)unstated · not in the catalog44set aside
- openai/gpt-oss-120bhigh41.80set aside
- qwen/qwen3-32bunstated4096.21
- gemini-exp-1206unstated · not in the catalog38.20set aside
- gemini-2.0-pro-exp-02-05unstated · not in the catalog35.6090.84
- grok-3-mini-betalow · not in the catalog34.7094.71
- o1-mini-2024-09-12unstated · not in the catalog32.90set aside
- openai/gpt-4.1-miniunstated32.40set aside
- claude-3-5-haiku-20241022unstated · not in the catalog2880.42
- chatgpt-4o-latest-(2025-02-15)unstated · not in the catalog27.10set aside
- qwq-32b-+-qwen-2.5-coder-instructunstated · not in the catalog26.20set aside
- openai/gpt-4o-2024-08-06unstated23.1082.76
- gemini-2.0-flash-expunstated · not in the catalog22.20set aside
- qwen-max-2025-01-25unstated · not in the catalog21.8085.11
- qwq-32bunstated · not in the catalog20.9093.18
- openai/gpt-4o-2024-11-20unstated18.2080.37
- gemini-2.0-flash-thinking-exp-01-21unstated · not in the catalog18.2092.14
- deepseek-chat-v2.5unstated · not in the catalog17.80set aside
- qwen2.5-coder-32b-instructunstated · not in the catalog16.40set aside
- meta-llama/llama-4-maverickunstated15.60set aside
- yi-lightningunstated · not in the catalog12.90set aside
- command-a-03-2025-qualityunstated · not in the catalog12set aside
- codestral-25.01unstated · not in the catalog11.10set aside
- openhands-lm-32b-v0.1unstated · not in the catalog10.20set aside
- openai/gpt-4.1-nanounstated8.90set aside
- qwen2.5-coder-32b-instructunstated · not in the catalog8set aside
- google/gemma-3-27b-itunstated4.9085.05
- openai/gpt-4o-mini-2024-07-18unstated3.6080.10
OTIS Mock AIME 2024-2025Epoch AI Benchmarking Hub-0.39 · 2.55 · 319 runs
319 runs scored on it, 266 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5high1119.29
- google/gemini-4-argonhigh1124.90
- google/gemini-4-carbonhigh1131.68
- google/gemini-4-hydrogenxhigh1125.83
- openai/gpt-5.6-solmax1123.35
- openai/gpt-6-astramax1132.41
- openai/gpt-6-solmax1124.00
- openai/gpt-6.1-solmax1130.00
- openai/gpt-6.1-terramax1130.85
- qwen/qwen3.8-max-0902xhigh1113.51
- claude-fable-5-1max · not in the catalog1125.12
- claude-opus-5-5max · not in the catalog1128.09
- claude-sonnet-5-5max · not in the catalog1122.13
- gemini-4-neonxhigh · not in the catalog1134.54
- gpt-5.5-pre-releasexhigh · not in the catalog1116.86
- gpt-5.5-pro-pre-releasexhigh · not in the catalog1117.38
- gpt-6.2-astramax · not in the catalog1135.84
- kimi-k3.1-pro-togethermax · not in the catalog1118.88
- anthropic/claude-fable-5max1.00122.58
- openai/gpt-5.6-terramax1.00119.24
- claude-fable-5-2max · not in the catalog1.00132.06
- deepseek/deepseek-v4.1-flash-1017max0.99116.25
- mistral-large-4-0-osshigh · not in the catalog0.99110.78
- muse-wonder-1.0max · not in the catalog0.99120.86
- qwen3.8-maxxhigh · not in the catalog0.99116.39
- meta/muse-spark-1.3xhigh0.99116.79
- doubao-seed-evolvinghigh · not in the catalog0.99114.76
- x-ai/grok-4.6xhigh0.99114.39
- anthropic/claude-opus-5max0.99122.24
- google/gemini-3.8-flashhigh0.99114.21
- openai/gpt-6-lunamax0.99113.99
- claude-haiku-5-5xhigh · not in the catalog0.99112.88
- deepseek/deepseek-v4-pro-0813max0.99113.60
- deepseek/deepseek-v4.1-flashmax0.98116.49
- mistralai/mistral-large-4-0high0.98110.63
- openai/gpt-5.6-lunamax0.98114.95
- claude-opus-4-8max · not in the catalog0.98116.89
- muse-spark-open-816b-a42bxhigh · not in the catalog0.98113.92
- x-ai/grok-4.7xhigh0.98113.32
- claude-opus-4-7xhigh · not in the catalog0.98113.14
- anthropic/claude-fable-5low0.98110.20
- anthropic/claude-opus-5unstated0.98115.94
- claude-opus-4-8low · not in the catalog0.98111.54
- gpt-5.4-2026-03-05high · not in the catalog0.98115.98
- x-ai/grok-4.5high0.98111.55
- x-ai/grok-4.6high0.98114.02
- google/gemini-3.7-flashhigh0.97116.01
- moonshotai/kimi-k3max0.97114.75
- claude-haiku-5-5high · not in the catalog0.97set aside
- deepseek/deepseek-v4-promax0.97107.02
- moonshotai/kimi-k2.6unstated0.96110.82
- doubao-seed-2.1-turbo-260628high · not in the catalog0.96108.84
- gpt-5.2-2025-12-11high · not in the catalog0.96110.74
- gpt-5.2-2025-12-11xhigh · not in the catalog0.96113.39
- google/gemini-3.1-pro-previewunstated0.96111.66
- deepseek/deepseek-v4-prohigh0.96108.20
- google/gemini-3-flash-previewhigh0.96113.70
- google/gemini-3.1-pro-previewhigh0.96113.89
- google/gemini-3.5-flashhigh0.96111.12
- moonshotai/kimi-k2.7-codeunstated0.96107.73
- openai/gpt-5.6-sollow0.96110.85
- qwen/qwen3.7-maxunstated0.96111.06
- doubao-seed-2.1-pro-260628high · not in the catalog0.96109.61
- gpt-5.4-2026-03-05medium · not in the catalog0.96112.04
- gpt-5.4-2026-03-05xhigh · not in the catalog0.95116.23
- anthropic/claude-sonnet-5xhigh0.95113.50
- deepseek/deepseek-v4-flash-0731max0.94111.14
- claude-opus-4-6-64kunstated · not in the catalog0.94111.92
- google/gemini-3.6-flashhigh0.94110.95
- doubao-seed-2.0-pro-260215high · not in the catalog0.94108.12
- gpt-5.2-2025-12-11medium · not in the catalog0.94108.75
- z-ai/glm-5.3-flashmax0.94108.45
- anthropic/claude-opus-5low0.93109.13
- moonshotai/kimi-k3high0.93112.20
- qwen/qwen3.6-plusunstated0.93106.49
- qwen/qwen3.7-plusunstated0.93107.80
- x-ai/grok-4.3high0.93107.39
- z-ai/glm-5.1unstated0.93109.27
- claude-opus-4-6-32kunstated · not in the catalog0.93111.03
- google/gemini-3-flash-previewunstated0.93108.16
- grok-4.20-0309-reasoningunstated · not in the catalog0.92108.33
- fireworks/kimi-k2p5unstated · not in the catalog0.92107.32
- gemini-3-pro-previewunstated · not in the catalog0.91110.74
- gpt-5-2025-08-07high · not in the catalog0.91108.61
- qwen/qwen3.6-27bunstated0.91103.72
- claude-opus-4-6max · not in the catalog0.91111.69
- doubao-seed-2.0-lite-260428high · not in the catalog0.91106.42
- qwen3.6-max-previewunstated · not in the catalog0.91106.75
- z-ai/glm-5.3max0.91112.54
- thinkingmachines/inkling-smallxhigh0.90106.54
- muse-sparkunstated · not in the catalog0.89110.41
- google/gemini-3.5-flashlow0.89110.83
- openai/gpt-5.6-terralow0.89106.93
- qwen/qwen3.5-397b-a17bunstated0.89104.40
- thinkingmachines/inklingxhigh0.89107.98
- gpt-5.4-mini-2026-03-17xhigh · not in the catalog0.89106.91
- openai/gpt-oss-120bhigh · not in the catalog0.89set aside
- gpt-5.1-2025-11-13high · not in the catalog0.89109.94
- deepseek-reasonerunstated · not in the catalog0.88set aside
- gpt-5.4-nano-2026-03-17high · not in the catalog0.88106.76
- gpt-5-2025-08-07medium · not in the catalog0.87108.12
- gpt-5.4-mini-2026-03-17high · not in the catalog0.87106.09
- qwen/qwen3-235b-a22b-thinking-2507unstated0.87104.10
- qwen/qwen3.6-35b-a3bunstated0.87102.04
- qwen/qwen3.7-flashunstated0.87102.16
- claude-opus-4-7max · not in the catalog0.87113.03
- gpt-5-mini-2025-08-07high · not in the catalog0.87104.57
- nemotron-3-ultraunstated · not in the catalog0.87105.62
- qwen3.5-plusunstated · not in the catalog0.87105.31
- z-ai/glm-5.2max0.86110.33
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.86104.05
- claude-sonnet-4-6-32kunstated · not in the catalog0.86set aside
- gpt-5.1-2025-11-13medium · not in the catalog0.86105.41
- openai/gpt-5.5low0.84107.86
- qwen/qwen3.6-flashunstated0.84100.95
- claude-opus-4-8none · not in the catalog0.84104.31
- gpt-5.4-2026-03-05low · not in the catalog0.84104.16
- o3-2025-04-16medium · not in the catalog0.84101.11
- qwen3.5-flashunstated · not in the catalog0.8499.58
- google/gemini-2.5-prounstated0.84103.73
- grok-4-0709unstated · not in the catalog0.84104.16
- o3-2025-04-16high · not in the catalog0.84102.96
- z-ai/glm-4.7unstated0.83104.07
- kimi-k2-thinking-turbounstated · not in the catalog0.83102.46
- google/gemini-3.6-flashlow0.82107.80
- google/gemma-4-26b-a4b-itminimal0.82set aside
- qwen/qwen3.5-397b-a17bnone0.82103.55
- claude-sonnet-4-6medium · not in the catalog0.82103.01
- claude-opus-4-5-20251101-16kunstated · not in the catalog0.82104.70
- o4-mini-2025-04-16high · not in the catalog0.82103.35
- gpt-5-nano-2025-08-07high · not in the catalog0.81100.45
- anthropic/claude-sonnet-5max0.80112.43
- google/gemini-3.1-flash-litehigh0.80102.13
- google/gemini-3.5-flashminimal0.80104.41
- google/gemini-3.6-flashminimal0.80105.55
- qwen/qwen3.7-plusnone0.80103.17
- z-ai/glm-5unstated0.8099.96
- gpt-5.2-2025-12-11low · not in the catalog0.79104.89
- gpt-5-mini-2025-08-07medium · not in the catalog0.78103.83
- qwen/qwen3.7-flashnone0.78100.54
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.78101.18
- claude-sonnet-4-5-20250929-59kunstated · not in the catalog0.78101.26
- grok-3-mini-betahigh · not in the catalog0.7897.03
- o3-mini-2025-01-31high · not in the catalog0.77102.45
- claude-sonnet-4-6high · not in the catalog0.76102.30
- z-ai/glm-5.2low0.76105.42
- gpt-5-nano-2025-08-07medium · not in the catalog0.7499.03
- google/gemma-4-31b-itminimal0.73set aside
- o1-2024-12-17high · not in the catalog0.7398.27
- o1-2024-12-17medium · not in the catalog0.7397.81
- o4-mini-2025-04-16medium · not in the catalog0.73101.93
- qwen3-max-2025-09-23unstated · not in the catalog0.73100.03
- gemini-2.5-flash-preview-04-17unstated · not in the catalog0.73set aside
- google/gemini-3.5-flash-litehigh0.7199.97
- minimax/minimax-m3unstated0.71104.58
- claude-sonnet-4-20250514-32kunstated · not in the catalog0.71set aside
- claude-sonnet-4-5-20250929-16kunstated · not in the catalog0.71set aside
- claude-sonnet-4-6max · not in the catalog0.71101.17
- gemini-2.5-flash-preview-05-20unstated · not in the catalog0.71set aside
- qwen3-30b-a3b-thinking-2507unstated · not in the catalog0.70set aside
- qwen/qwen3.5-35b-a3bnone0.70set aside
- seed-1.6unstated · not in the catalog0.6997.70
- moonshotai/kimi-k3low0.69100.12
- openai/gpt-5.6-solnone0.69100.98
- qwen/qwen3.6-35b-a3bnone0.69100.22
- claude-opus-4-1-20250805-27kunstated · not in the catalog0.6997.29
- claude-sonnet-4-20250514-59kunstated · not in the catalog0.6999.17
- gpt-5.4-nano-2026-03-17low · not in the catalog0.6998.40
- gpt-5.5-instantunstated · not in the catalog0.6899.89
- seed-oss-36b-instructunstated · not in the catalog0.68set aside
- qwen/qwen3-32bunstated0.6796.21
- openai/gpt-5.6-lunalow0.67101.92
- qwen/qwen3.6-27bnone0.67100.32
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.6796.01
- deepseek/deepseek-r1-0528unstated0.6698.81
- qwen/qwen3-14bunstated0.66set aside
- openai/gpt-oss-20bmedium0.65set aside
- claude-opus-4-1-20250805-16kunstated · not in the catalog0.6496.39
- claude-opus-4-20250514-27kunstated · not in the catalog0.6497.69
- gpt-5.1-2025-11-13low · not in the catalog0.64100.64
- o3-mini-2025-01-31medium · not in the catalog0.6497.94
- qwen/qwen3-30b-a3bunstated0.63set aside
- qwen/qwen3-30b-a3b-instruct-2507unstated0.62set aside
- gpt-5.2-2025-12-11none · not in the catalog0.6296.53
- grok-3-mini-betalow · not in the catalog0.6294.71
- qwen/qwen3.5-9bnone0.62set aside
- google/gemini-3.5-flash-litelow0.6098.68
- claude-opus-4-20250514-16kunstated · not in the catalog0.60set aside
- o3-2025-04-16low · not in the catalog0.6099.37
- qwq-32bunstated · not in the catalog0.5993.18
- z-ai/glm-4.7-flashunstated0.58set aside
- openai/gpt-5.5none0.5897.11
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog0.5895.18
- gemini-2.0-flash-thinking-exp-01-21unstated · not in the catalog0.5892.14
- gpt-5.4-2026-03-05none · not in the catalog0.5896.15
- o4-mini-2025-04-16low · not in the catalog0.58100.61
- qwen3-8bunstated · not in the catalog0.56set aside
- qwen3.5-4bnone · not in the catalog0.56set aside
- deepseek-r1-distill-qwen-32bunstated · not in the catalog0.56set aside
- gpt-5-mini-2025-08-07minimal · not in the catalog0.5693.86
- grok-3-betaunstated · not in the catalog0.5694.65
- qwen/qwen3.5-35b-a3bunstated0.54set aside
- openai/gpt-5.6-terranone0.5395.57
- claude-3-7-sonnet-20250219-32kunstated · not in the catalog0.5394.40
- claude-sonnet-4-20250514-16kunstated · not in the catalog0.53set aside
- deepseek-r1unstated · not in the catalog0.5396.08
- o1-2024-12-17low · not in the catalog0.5394.71
- qwen3-4b-instruct-2507unstated · not in the catalog0.52set aside
- deepseek-r1-distill-llama-70bunstated · not in the catalog0.5193.55
- google/gemini-3.5-flash-liteminimal0.5194.95
- openai/gpt-oss-20bhigh0.51set aside
- deepseek-r1-distill-qwen-14bunstated · not in the catalog0.5192.30
- deepseek/deepseek-chatunstated0.49set aside
- claude-opus-4-5-20251101unstated · not in the catalog0.4899.92
- o1-mini-2024-09-12high · not in the catalog0.4792.53
- deepseek/deepseek-v4-pronone0.4794.28
- claude-3-7-sonnet-20250219-16kunstated · not in the catalog0.4793.20
- gpt-5-2025-08-07minimal · not in the catalog0.4795.14
- gpt-5-nano-2025-08-07low · not in the catalog0.4792.74
- gpt-5.4-nano-2026-03-17none · not in the catalog0.4791.33
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.4593.48
- o1-mini-2024-09-12medium · not in the catalog0.4591.51
- google/gemini-3.1-flash-litelow0.4494.94
- o3-mini-2025-01-31low · not in the catalog0.4491.91
- deepseek-r1-0528-qwen3-8bunstated · not in the catalog0.44set aside
- qwen/qwen3.5-9bunstated0.43set aside
- claude-opus-4-20250514unstated · not in the catalog0.4292.48
- openai/gpt-oss-20blow0.40set aside
- openai/gpt-5.6-lunanone0.4095.93
- claude-opus-4-1-20250805unstated · not in the catalog0.4093.81
- gpt-4.1-2025-04-14unstated · not in the catalog0.3891.64
- google/gemini-3.1-flash-liteminimal0.3893.68
- deepseek-v3-0324unstated · not in the catalog0.3890.12
- gpt-4.5-preview-2025-02-27unstated · not in the catalog0.3890.75
- gpt-5.1-2025-11-13none · not in the catalog0.3891.90
- claude-haiku-4-5-20251001unstated · not in the catalog0.3690.99
- claude-sonnet-4-5-20250929unstated · not in the catalog0.3693.21
- gpt-5-nano-2025-08-07minimal · not in the catalog0.3688.08
- mistral-medium-2505unstated · not in the catalog0.3289.41
- gemini-2.0-flash-001unstated · not in the catalog0.3189.75
- o1-preview-2024-09-12unstated · not in the catalog0.3189.08
- mistral-small-3.2-2506unstated · not in the catalog0.30set aside
- magistral-small-2506unstated · not in the catalog0.3088.63
- claude-sonnet-4-20250514unstated · not in the catalog0.2990.16
- gpt-4.1-nano-2025-04-14unstated · not in the catalog0.2986.22
- z-ai/glm-5.2none0.2996.22
- magistral-small-2509unstated · not in the catalog0.28set aside
- minimax/minimax-m3none0.2790.84
- gpt-5.4-mini-2026-03-17none · not in the catalog0.2790.48
- qwen/qwen3-14bnone0.26set aside
- qwen/qwen3-30b-a3bnone0.26set aside
- z-ai/glm-4.7-flashnone0.25set aside
- qwen/qwen3-32bnone0.23set aside
- gemini-1.5-pro-002unstated · not in the catalog0.2386.80
- google/gemma-3-27b-itunstated0.2385.05
- qwen3-8bnone · not in the catalog0.22set aside
- claude-3-7-sonnet-20250219unstated · not in the catalog0.2287.22
- deepseek-r1-distill-qwen-1.5bunstated · not in the catalog0.21set aside
- llama-4-maverick-17b-128e-instruct-fp8unstated · not in the catalog0.2187.21
- qwen-plus-2025-01-25unstated · not in the catalog0.1884.76
- google/gemma-3-12b-itunstated0.17set aside
- gemini-1.5-flash-002unstated · not in the catalog0.1683.86
- qwen-max-2025-01-25unstated · not in the catalog0.1685.11
- deepseek-v3unstated · not in the catalog0.1684.96
- microsoft/phi-4unstated0.1484.64
- grok-2-1212unstated · not in the catalog0.1283.89
- llama-3.1-405b-instructunstated · not in the catalog0.1081.49
- mistralai/mistral-large-2407unstated0.0880.40
- claude-3-5-sonnet-20241022unstated · not in the catalog0.0885.31
- qwen2.5-72b-instructunstated · not in the catalog0.0883.11
- qwen3-1.7bnone · not in the catalog0.08set aside
- llama-4-scout-17b-16e-instructunstated · not in the catalog0.0882.99
- mistral-large-2411unstated · not in the catalog0.0881.24
- google/gemma-3-4b-itunstated0.07set aside
- qwen2.5-32b-instructunstated · not in the catalog0.0781.76
- openai/gpt-4o-mini-2024-07-18unstated0.0780.10
- gemini-1.5-pro-001unstated · not in the catalog0.0779.25
- gpt-4-turbo-2024-04-09unstated · not in the catalog0.0780.07
- mistral-small-3-2501unstated · not in the catalog0.07set aside
- claude-3-5-sonnet-20240620unstated · not in the catalog0.0781.59
- openai/gpt-4o-2024-08-06unstated0.0682.76
- openai/gpt-4o-2024-05-13unstated0.0681.01
- openai/gpt-4o-2024-11-20unstated0.0680.37
- qwen-turbo-2024-11-01unstated · not in the catalog0.0681.30
- mistral-small-2503unstated · not in the catalog0.0680.18
- mistral-small-2501unstated · not in the catalog0.0579.62
- meta-llama/llama-3.3-70b-instructunstated0.0579.22
- claude-3-opus-20240229unstated · not in the catalog0.0578.88
- gemini-1.5-flash-8b-001unstated · not in the catalog0.05set aside
- llama-3.1-tulu-3-70b-dpounstated · not in the catalog0.0479.20
- claude-3-5-haiku-20241022unstated · not in the catalog0.0480.42
- meta-llama-3-70b-instructunstated · not in the catalog0.0474.91
- gemini-1.5-flash-001unstated · not in the catalog0.0475.35
- mistral-small-3.1-2503unstated · not in the catalog0.04set aside
- meta-llama/llama-3.1-70b-instructunstated0.0477.86
- granite-4.0-microunstated · not in the catalog0.03set aside
- llama-3.2-90b-vision-instructunstated · not in the catalog0.0377.93
- claude-2.0unstated · not in the catalog0.0370.91
- claude-3-sonnet-20240229unstated · not in the catalog0.0373.46
- hermes-2-theta-llama-3-70bunstated · not in the catalog0.0374.21
- qwen2.5-7b-instructunstated · not in the catalog0.03set aside
- gpt-3.5-turbo-0125unstated · not in the catalog0.0269.92
- claude-2.1unstated · not in the catalog0.02set aside
- meta-llama-3-8b-instructunstated · not in the catalog0.0267.01
- mistral-large-2402unstated · not in the catalog0.0274.62
- claude-3-haiku-20240307unstated · not in the catalog0.0271.87
- meta-llama/llama-3.1-8b-instructunstated0.0272.89
- google/gemma-2-27b-itunstated0.0175.03
- gemini-1.0-pro-001unstated · not in the catalog0.0170.18
- gemma-3-1b-itunstated · not in the catalog0.01set aside
- gpt-4-0613unstated · not in the catalog0.0174.56
- deepseek-llm-67b-chatunstated · not in the catalog0.0166.23
- granite-4.0-1bunstated · not in the catalog0.01set aside
- meta-llama/llama-3.2-1b-instructunstated0.01set aside
- gemma-2-9b-itunstated · not in the catalog0.0172.14
- gpt-4-0314unstated · not in the catalog0.01set aside
- granite-4.0-350munstated · not in the catalog0.00set aside
- mistral-7b-instruct-v0.3unstated · not in the catalog0.0060.84
- llama-2-70b-chat-hfunstated · not in the catalog064.40
simplifyLiveBench-0.56 · 0.37 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- google/gemini-3.6-flashhigh75.70110.95
- google/gemini-3.8-flashhigh75.40114.21
- google/gemini-3.7-flashhigh74.62116.01
- meta/muse-spark-1.3xhigh72.52116.79
- claude-sonnet-5-5xhigh · not in the catalog72.37119.88
- nvidia/nemotron-3-ultra-550b-a55bunstated72.28105.72
- anthropic/claude-fable-5max72.03122.58
- openai/gpt-6.1-solmax71.97130.00
- google/gemini-3.1-pro-previewhigh71.05113.89
- openai/gpt-6-astramax70.53132.41
- openai/gpt-5.6-solmax70.52123.35
- qwen3.8-flash-nextunstated · not in the catalog70.25113.12
- claude-fable-5-1max · not in the catalog70.17125.12
- x-ai/grok-4.5unstated70.17114.74
- openai/gpt-5.4xhigh70119.02
- x-ai/grok-4.7xhigh69.97113.32
- x-ai/grok-4.6unstated69.33117.62
- openai/gpt-6-solmax69.28124.00
- google/gemini-3.5-flashhigh69.10111.12
- meta/muse-spark-1.2xhigh68.73117.04
- openai/gpt-6.1-solxhigh68.17124.64
- google/gemini-3.5-flash-litehigh68.0299.97
- qwen3.8-maxunstated · not in the catalog67.35118.08
- deepseek/deepseek-v4.1-flashmax67.07116.49
- muse-wonder-1.0max · not in the catalog67.05120.86
- qwen/qwen3.8-27bunstated66.70112.72
- smaug-agenticunstated · not in the catalog66.62116.30
- moonshotai/kimi-k3unstated66.37116.54
- openai/gpt-5.5xhigh66.32120.29
- mistral-large-4high · not in the catalog66.18109.81
- z-ai/glm-5.3unstated65.85112.92
- meta/muse-spark-1.1xhigh65.70114.21
- claude-opus-4-7xhigh · not in the catalog65.50113.14
- smaug-miniunstated · not in the catalog65.38114.73
- deepseek/deepseek-v4-flash-vision-expunstated65.07116.36
- qwen/qwen3.7-maxunstated64.08111.06
- smaug-flashunstated · not in the catalog63.13116.63
- claude-opus-5-5xhigh · not in the catalog63125.14
- claude-sonnet-4-6-thinking-automedium · not in the catalog62.93114.06
- claude-haiku-5-5xhigh · not in the catalog62.78112.88
- openai/gpt-5.4-nanoxhigh62.75110.30
- x-ai/grok-build-0.1unstated62.60106.25
- claude-opus-4-8max · not in the catalog62116.89
- anthropic/claude-opus-5max61.58122.24
- deepseek/deepseek-v4-pro-0813unstated61.57118.60
- openai/gpt-5.6-terramax61.43119.24
- kimi-k2.6-thinkingunstated · not in the catalog61.42107.26
- anthropic/claude-sonnet-5xhigh60.65113.50
- openai/gpt-5.2-codexunstated60.57110.02
- thinkingmachines/inklingxhigh60.50107.98
- claude-opus-4-6-thinking-autohigh · not in the catalog59.15113.04
- openai/gpt-5.6-lunamax58.67114.95
- claude-opus-5-5max · not in the catalog58.67128.09
- deepseek/deepseek-v4-flash-0731unstated58.37113.47
- x-ai/grok-4.3unstated57.60103.07
- minimax/minimax-m3unstated56.62104.58
- deepseek/deepseek-v4-prounstated55.93111.64
- z-ai/glm-5.2unstated55.67109.40
- openai/gpt-6-lunamax55.43113.99
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog54.95110.65
- ox-alphamax · not in the catalog54.48106.05
- gpt-5.2-2025-12-11high · not in the catalog54.08110.74
- openai/gpt-5.4-minixhigh54.07104.22
- deepseek/deepseek-v4-flashunstated54.05102.96
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog52.6887.31
- qwen/qwen3.6-plusunstated51.57106.49
- claude-sonnet-5-5max · not in the catalog50.85122.13
- qwen/qwen3.6-27bunstated50.73103.72
- claude-haiku-5-5max · not in the catalog50.25113.88
- moonshotai/kimi-k2.7-codeunstated49.75107.73
- z-ai/glm-5.3-flashunstated46.95108.42
connectionsLiveBench-0.62 · 2.14 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- deepseek/deepseek-v4-pro-0813unstated100118.60
- deepseek/deepseek-v4.1-flashmax100116.49
- google/gemini-3.1-pro-previewhigh100113.89
- google/gemini-3.5-flashhigh100111.12
- google/gemini-3.6-flashhigh100110.95
- google/gemini-3.7-flashhigh100116.01
- google/gemini-3.8-flashhigh100114.21
- meta/muse-spark-1.2xhigh100117.04
- moonshotai/kimi-k3unstated100116.54
- openai/gpt-5.4xhigh100119.02
- openai/gpt-5.5xhigh100120.29
- openai/gpt-5.6-solmax100123.35
- openai/gpt-6-astramax100132.41
- openai/gpt-6-lunamax100113.99
- openai/gpt-6-solmax100124.00
- openai/gpt-6.1-solmax100130.00
- openai/gpt-6.1-solxhigh100124.64
- smaug-agenticunstated · not in the catalog100116.30
- x-ai/grok-4.6unstated100117.62
- anthropic/claude-fable-5max99.33122.58
- anthropic/claude-opus-5max99.33122.24
- deepseek/deepseek-v4-flash-vision-expunstated99.33116.36
- claude-fable-5-1max · not in the catalog99.33125.12
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog99.33110.65
- claude-opus-4-6-thinking-autohigh · not in the catalog99.33113.04
- claude-opus-4-8max · not in the catalog99.33116.89
- claude-opus-5-5max · not in the catalog99.33128.09
- claude-opus-5-5xhigh · not in the catalog99.33125.14
- claude-sonnet-4-6-thinking-automedium · not in the catalog99.33114.06
- claude-sonnet-5-5max · not in the catalog99.33122.13
- smaug-flashunstated · not in the catalog99.33116.63
- gpt-5.2-2025-12-11high · not in the catalog99110.74
- openai/gpt-5.4-minixhigh98.98104.22
- claude-sonnet-5-5xhigh · not in the catalog98.50119.88
- deepseek/deepseek-v4-prounstated98111.64
- meta/muse-spark-1.1xhigh98114.21
- moonshotai/kimi-k2.7-codeunstated98107.73
- x-ai/grok-4.5unstated98114.74
- x-ai/grok-4.7xhigh98113.32
- deepseek/deepseek-v4-flash-0731unstated97.33113.47
- openai/gpt-5.6-lunamax96.50114.95
- qwen/qwen3.7-maxunstated96.50111.06
- minimax/minimax-m3unstated96.33104.58
- claude-haiku-5-5max · not in the catalog96113.88
- muse-wonder-1.0max · not in the catalog96120.86
- anthropic/claude-sonnet-5xhigh95.50113.50
- z-ai/glm-5.3unstated95.33112.92
- openai/gpt-5.2-codexunstated95110.02
- qwen3.8-maxunstated · not in the catalog94.50118.08
- openai/gpt-5.4-nanoxhigh94.39110.30
- meta/muse-spark-1.3xhigh94116.79
- x-ai/grok-4.3unstated94103.07
- z-ai/glm-5.2unstated94109.40
- claude-haiku-5-5xhigh · not in the catalog93.83112.88
- z-ai/glm-5.3-flashunstated93.67108.42
- openai/gpt-5.6-terramax93119.24
- claude-opus-4-7xhigh · not in the catalog92.33113.14
- google/gemini-3.5-flash-litehigh9299.97
- smaug-miniunstated · not in the catalog91.67114.73
- thinkingmachines/inklingxhigh91.17107.98
- qwen/qwen3.6-plusunstated90.50106.49
- qwen/qwen3.8-27bunstated90.50112.72
- qwen3.8-flash-nextunstated · not in the catalog90.17113.12
- deepseek/deepseek-v4-flashunstated89.50102.96
- kimi-k2.6-thinkingunstated · not in the catalog89.33107.26
- nvidia/nemotron-3-ultra-550b-a55bunstated86.67105.72
- x-ai/grok-build-0.1unstated84106.25
- qwen/qwen3.6-27bunstated77.17103.72
- ox-alphamax · not in the catalog71.33106.05
- mistral-large-4high · not in the catalog48109.81
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog3487.31
theory_of_mindLiveBench-0.80 · 0.75 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-5.4xhigh88.46119.02
- z-ai/glm-5.3unstated88.46112.92
- deepseek/deepseek-v4-flash-0731unstated86.54113.47
- openai/gpt-5.6-terramax86.54119.24
- openai/gpt-6.1-solmax86.54130.00
- openai/gpt-6.1-solxhigh86.54124.64
- claude-haiku-5-5max · not in the catalog86.54113.88
- claude-sonnet-5-5max · not in the catalog86.54122.13
- muse-wonder-1.0max · not in the catalog86.54120.86
- x-ai/grok-4.6unstated86.54117.62
- anthropic/claude-fable-5max84.61122.58
- deepseek/deepseek-v4-pro-0813unstated84.61118.60
- meta/muse-spark-1.3xhigh84.61116.79
- openai/gpt-5.5xhigh84.61120.29
- openai/gpt-5.6-solmax84.61123.35
- openai/gpt-6-astramax84.61132.41
- openai/gpt-6-solmax84.61124.00
- claude-opus-5-5max · not in the catalog84.61128.09
- claude-opus-5-5xhigh · not in the catalog84.61125.14
- x-ai/grok-4.7xhigh84.61113.32
- google/gemini-3.7-flashhigh82.69116.01
- moonshotai/kimi-k3unstated82.69116.54
- claude-opus-4-6-thinking-autohigh · not in the catalog82.69113.04
- x-ai/grok-4.5unstated82.69114.74
- anthropic/claude-sonnet-5xhigh80.77113.50
- deepseek/deepseek-v4-prounstated80.77111.64
- deepseek/deepseek-v4.1-flashmax80.77116.49
- google/gemini-3.1-pro-previewhigh80.77113.89
- google/gemini-3.5-flashhigh80.77111.12
- meta/muse-spark-1.2xhigh80.77117.04
- openai/gpt-5.4-nanoxhigh80.77110.30
- claude-fable-5-1max · not in the catalog80.77125.12
- claude-opus-4-7xhigh · not in the catalog80.77113.14
- claude-opus-4-8max · not in the catalog80.77116.89
- mistral-large-4high · not in the catalog80.77109.81
- qwen3.8-flash-nextunstated · not in the catalog80.77113.12
- anthropic/claude-opus-5max78.85122.24
- deepseek/deepseek-v4-flash-vision-expunstated78.85116.36
- google/gemini-3.6-flashhigh78.85110.95
- openai/gpt-5.2-codexunstated78.85110.02
- qwen/qwen3.7-maxunstated78.85111.06
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog78.85110.65
- gpt-5.2-2025-12-11high · not in the catalog78.85110.74
- ox-alphamax · not in the catalog78.85106.05
- qwen3.8-maxunstated · not in the catalog78.85118.08
- smaug-agenticunstated · not in the catalog78.85116.30
- google/gemini-3.8-flashhigh76.92114.21
- meta/muse-spark-1.1xhigh76.92114.21
- minimax/minimax-m3unstated76.92104.58
- openai/gpt-5.4-minixhigh76.92104.22
- claude-haiku-5-5xhigh · not in the catalog76.92112.88
- smaug-flashunstated · not in the catalog76.92116.63
- kimi-k2.6-thinkingunstated · not in the catalog75107.26
- z-ai/glm-5.2unstated75109.40
- deepseek/deepseek-v4-flashunstated73.08102.96
- openai/gpt-5.6-lunamax73.08114.95
- openai/gpt-6-lunamax73.08113.99
- claude-sonnet-4-6-thinking-automedium · not in the catalog73.08114.06
- z-ai/glm-5.3-flashunstated73.08108.42
- moonshotai/kimi-k2.7-codeunstated69.23107.73
- claude-sonnet-5-5xhigh · not in the catalog69.23119.88
- smaug-miniunstated · not in the catalog69.23114.73
- x-ai/grok-build-0.1unstated69.23106.25
- qwen/qwen3.6-plusunstated67.31106.49
- qwen/qwen3.6-27bunstated65.39103.72
- thinkingmachines/inklingxhigh65.39107.98
- nvidia/nemotron-3-ultra-550b-a55bunstated61.54105.72
- x-ai/grok-4.3unstated61.54103.07
- qwen/qwen3.8-27bunstated59.62112.72
- google/gemini-3.5-flash-litehigh5099.97
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog44.2387.31
code_completionLiveBench-0.87 · 0.68 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- openai/gpt-5.2-codexunstated86.96110.02
- openai/gpt-5.6-lunamax86.96114.95
- claude-opus-5-5max · not in the catalog86.96128.09
- claude-opus-5-5xhigh · not in the catalog86.96125.14
- claude-sonnet-5-5max · not in the catalog86.96122.13
- openai/gpt-5.6-solmax84.78123.35
- claude-opus-4-8max · not in the catalog84.78116.89
- claude-sonnet-5-5xhigh · not in the catalog84.78119.88
- anthropic/claude-opus-5max82.61122.24
- deepseek/deepseek-v4.1-flashmax82.61116.49
- moonshotai/kimi-k3unstated82.61116.54
- openai/gpt-5.5xhigh82.61120.29
- openai/gpt-6.1-solxhigh82.61124.64
- claude-fable-5-1max · not in the catalog82.61125.12
- anthropic/claude-fable-5max80.44122.58
- meta/muse-spark-1.2xhigh80.44117.04
- meta/muse-spark-1.3xhigh80.44116.79
- openai/gpt-5.4xhigh80.44119.02
- openai/gpt-5.6-terramax80.44119.24
- openai/gpt-6-astramax80.44132.41
- openai/gpt-6-lunamax80.44113.99
- openai/gpt-6-solmax80.44124.00
- openai/gpt-6.1-solmax80.44130.00
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog80.44110.65
- smaug-agenticunstated · not in the catalog80.44116.30
- smaug-miniunstated · not in the catalog80.44114.73
- z-ai/glm-5.2unstated80.44109.40
- z-ai/glm-5.3unstated80.44112.92
- z-ai/glm-5.3-flashunstated80.44108.42
- anthropic/claude-sonnet-5xhigh78.26113.50
- deepseek/deepseek-v4-pro-0813unstated78.26118.60
- google/gemini-3.1-pro-previewhigh78.26113.89
- google/gemini-3.6-flashhigh78.26110.95
- meta/muse-spark-1.1xhigh78.26114.21
- claude-haiku-5-5max · not in the catalog78.26113.88
- claude-opus-4-7xhigh · not in the catalog78.26113.14
- claude-sonnet-4-6-thinking-automedium · not in the catalog78.26114.06
- kimi-k2.6-thinkingunstated · not in the catalog78.26107.26
- mistral-large-4high · not in the catalog78.26109.81
- muse-wonder-1.0max · not in the catalog78.26120.86
- ox-alphamax · not in the catalog78.26106.05
- x-ai/grok-4.7xhigh78.26113.32
- google/gemini-3.5-flashhigh76.09111.12
- google/gemini-3.5-flash-litehigh76.0999.97
- google/gemini-3.7-flashhigh76.09116.01
- moonshotai/kimi-k2.7-codeunstated76.09107.73
- qwen/qwen3.6-plusunstated76.09106.49
- claude-opus-4-6-thinking-autohigh · not in the catalog76.09113.04
- gpt-5.2-2025-12-11high · not in the catalog76.09110.74
- qwen3.8-flash-nextunstated · not in the catalog76.09113.12
- x-ai/grok-4.6unstated76.09117.62
- openai/gpt-5.4-minixhigh75.56104.22
- deepseek/deepseek-v4-flash-0731unstated73.91113.47
- qwen/qwen3.8-27bunstated73.91112.72
- claude-haiku-5-5xhigh · not in the catalog73.91112.88
- qwen3.8-maxunstated · not in the catalog73.91118.08
- smaug-flashunstated · not in the catalog73.91116.63
- google/gemini-3.8-flashhigh71.74114.21
- qwen/qwen3.6-27bunstated71.74103.72
- deepseek/deepseek-v4-prounstated69.56111.64
- nvidia/nemotron-3-ultra-550b-a55bunstated69.56105.72
- qwen/qwen3.7-maxunstated69.56111.06
- x-ai/grok-4.5unstated69.56114.74
- deepseek/deepseek-v4-flash-vision-expunstated67.39116.36
- minimax/minimax-m3unstated67.39104.58
- openai/gpt-5.4-nanoxhigh67.39110.30
- thinkingmachines/inklingxhigh67.39107.98
- x-ai/grok-build-0.1unstated67.39106.25
- deepseek/deepseek-v4-flashunstated65.22102.96
- x-ai/grok-4.3unstated65.22103.07
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog23.9187.31
paraphraseLiveBench-0.94 · 0.36 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- google/gemini-3.8-flashhigh80114.21
- google/gemini-3.1-pro-previewhigh79.87113.89
- qwen3.8-flash-nextunstated · not in the catalog79.47113.12
- meta/muse-spark-1.3xhigh78.62116.79
- openai/gpt-6-astramax78.25132.41
- google/gemini-3.7-flashhigh78.07116.01
- x-ai/grok-4.7xhigh78.05113.32
- anthropic/claude-fable-5max77.65122.58
- nvidia/nemotron-3-ultra-550b-a55bunstated77.38105.72
- smaug-miniunstated · not in the catalog76.18114.73
- google/gemini-3.5-flashhigh75.52111.12
- qwen/qwen3.8-27bunstated74.90112.72
- moonshotai/kimi-k3unstated74.37116.54
- google/gemini-3.6-flashhigh74.32110.95
- meta/muse-spark-1.2xhigh74.30117.04
- muse-wonder-1.0max · not in the catalog74.07120.86
- qwen3.8-maxunstated · not in the catalog73.32118.08
- deepseek/deepseek-v4-flash-vision-expunstated72.83116.36
- openai/gpt-6.1-solmax72.58130.00
- qwen/qwen3.7-maxunstated72111.06
- x-ai/grok-4.6unstated71.78117.62
- openai/gpt-5.5xhigh71.73120.29
- openai/gpt-5.6-solmax71.52123.35
- claude-fable-5-1max · not in the catalog71.52125.12
- openai/gpt-6.1-solxhigh71.03124.64
- thinkingmachines/inklingxhigh70.85107.98
- deepseek/deepseek-v4.1-flashmax70.73116.49
- openai/gpt-6-solmax70.08124.00
- meta/muse-spark-1.1xhigh69.40114.21
- x-ai/grok-4.5unstated69.22114.74
- deepseek/deepseek-v4-pro-0813unstated68.63118.60
- smaug-agenticunstated · not in the catalog68.63116.30
- claude-opus-4-8max · not in the catalog67.87116.89
- openai/gpt-5.2-codexunstated67.05110.02
- openai/gpt-5.4-nanoxhigh66.65110.30
- claude-opus-4-7xhigh · not in the catalog65.77113.14
- anthropic/claude-opus-5max65.73122.24
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog65.67110.65
- smaug-flashunstated · not in the catalog65.57116.63
- google/gemini-3.5-flash-litehigh6599.97
- claude-opus-5-5xhigh · not in the catalog64.68125.14
- z-ai/glm-5.3unstated64.68112.92
- x-ai/grok-4.3unstated64.38103.07
- claude-opus-5-5max · not in the catalog64.27128.09
- gpt-5.2-2025-12-11high · not in the catalog64.10110.74
- mistral-large-4high · not in the catalog63.97109.81
- openai/gpt-5.4xhigh63.43119.02
- claude-opus-4-6-thinking-autohigh · not in the catalog62.82113.04
- claude-haiku-5-5xhigh · not in the catalog62.50112.88
- deepseek/deepseek-v4-flashunstated62.38102.96
- claude-sonnet-5-5xhigh · not in the catalog62.28119.88
- z-ai/glm-5.2unstated62.25109.40
- deepseek/deepseek-v4-flash-0731unstated61.70113.47
- kimi-k2.6-thinkingunstated · not in the catalog61.63107.26
- anthropic/claude-sonnet-5xhigh61.25113.50
- qwen/qwen3.6-plusunstated60.62106.49
- claude-sonnet-4-6-thinking-automedium · not in the catalog59.40114.06
- openai/gpt-5.4-minixhigh58.88104.22
- x-ai/grok-build-0.1unstated58.38106.25
- ox-alphamax · not in the catalog57.47106.05
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog57.3287.31
- openai/gpt-5.6-terramax57.03119.24
- deepseek/deepseek-v4-prounstated56.22111.64
- minimax/minimax-m3unstated55.93104.58
- claude-haiku-5-5max · not in the catalog50.22113.88
- openai/gpt-6-lunamax49.70113.99
- qwen/qwen3.6-27bunstated49.43103.72
- moonshotai/kimi-k2.7-codeunstated48.53107.73
- claude-sonnet-5-5max · not in the catalog48.22122.13
- z-ai/glm-5.3-flashunstated47.98108.42
- openai/gpt-5.6-lunamax47.82114.95
summarizeLiveBench-1.02 · 0.44 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- google/gemini-3.1-pro-previewhigh88.23113.89
- google/gemini-3.8-flashhigh87.50114.21
- google/gemini-3.7-flashhigh84.77116.01
- qwen/qwen3.7-maxunstated83.80111.06
- meta/muse-spark-1.3xhigh83.40116.79
- google/gemini-3.5-flashhigh82.58111.12
- anthropic/claude-fable-5max81.70122.58
- openai/gpt-6.1-solmax81.25130.00
- meta/muse-spark-1.2xhigh80.62117.04
- openai/gpt-6-astramax79.55132.41
- qwen3.8-maxunstated · not in the catalog79.15118.08
- smaug-flashunstated · not in the catalog78.35116.63
- x-ai/grok-4.7xhigh78.17113.32
- claude-opus-4-8max · not in the catalog77.67116.89
- qwen3.8-flash-nextunstated · not in the catalog77.35113.12
- muse-wonder-1.0max · not in the catalog76.83120.86
- smaug-miniunstated · not in the catalog76.10114.73
- qwen/qwen3.8-27bunstated75.93112.72
- google/gemini-3.6-flashhigh75.85110.95
- openai/gpt-6.1-solxhigh75.53124.64
- nvidia/nemotron-3-ultra-550b-a55bunstated74.88105.72
- thinkingmachines/inklingxhigh74.85107.98
- z-ai/glm-5.3unstated74.55112.92
- meta/muse-spark-1.1xhigh74.30114.21
- openai/gpt-5.6-terramax73.17119.24
- x-ai/grok-4.5unstated73.15114.74
- openai/gpt-5.5xhigh72.32120.29
- claude-fable-5-1max · not in the catalog72.17125.12
- deepseek/deepseek-v4-flash-vision-expunstated72.10116.36
- smaug-agenticunstated · not in the catalog72.05116.30
- openai/gpt-5.4-nanoxhigh71.77110.30
- deepseek/deepseek-v4-flash-0731unstated71.50113.47
- anthropic/claude-sonnet-5xhigh71.22113.50
- claude-sonnet-5-5xhigh · not in the catalog70.33119.88
- deepseek/deepseek-v4.1-flashmax69.65116.49
- deepseek/deepseek-v4-prounstated69.50111.64
- openai/gpt-5.6-solmax69.43123.35
- moonshotai/kimi-k3unstated69.35116.54
- claude-opus-5-5xhigh · not in the catalog69.33125.14
- claude-opus-5-5max · not in the catalog69.10128.09
- deepseek/deepseek-v4-flashunstated68.85102.96
- google/gemini-3.5-flash-litehigh68.4799.97
- openai/gpt-5.4xhigh68.42119.02
- z-ai/glm-5.2unstated68.37109.40
- x-ai/grok-4.6unstated68.08117.62
- openai/gpt-6-solmax68124.00
- kimi-k2.6-thinkingunstated · not in the catalog67.50107.26
- x-ai/grok-build-0.1unstated67.37106.25
- claude-haiku-5-5xhigh · not in the catalog66.98112.88
- anthropic/claude-opus-5max66.43122.24
- claude-opus-4-7xhigh · not in the catalog66.28113.14
- deepseek/deepseek-v4-pro-0813unstated65.67118.60
- claude-opus-4-6-thinking-autohigh · not in the catalog65.15113.04
- openai/gpt-5.6-lunamax64.93114.95
- qwen/qwen3.6-plusunstated64.32106.49
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog63.72110.65
- openai/gpt-5.2-codexunstated63.38110.02
- gpt-5.2-2025-12-11high · not in the catalog63.20110.74
- claude-sonnet-4-6-thinking-automedium · not in the catalog61.65114.06
- mistral-large-4high · not in the catalog61.63109.81
- openai/gpt-5.4-minixhigh59.63104.22
- x-ai/grok-4.3unstated58.42103.07
- moonshotai/kimi-k2.7-codeunstated58.32107.73
- ox-alphamax · not in the catalog58.18106.05
- claude-sonnet-5-5max · not in the catalog58.17122.13
- z-ai/glm-5.3-flashunstated56.47108.42
- openai/gpt-6-lunamax55.32113.99
- minimax/minimax-m3unstated53.93104.58
- claude-haiku-5-5max · not in the catalog53.63113.88
- qwen/qwen3.6-27bunstated52.62103.72
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog50.4287.31
SWE-Bench verifiedEpoch AI Benchmarking Hub-1.06 · 0.66 · 33 runs
33 runs scored on it, 29 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-opus-4-7max · not in the catalog0.83113.03
- gpt-5.5-pre-releasexhigh · not in the catalog0.81116.86
- google/gemini-3.5-flashhigh0.79111.12
- claude-opus-4-6unstated · not in the catalog0.79110.08
- z-ai/glm-5.2max0.79110.33
- deepseek/deepseek-v4-promax0.78107.02
- qwen/qwen3.7-maxunstated0.77111.06
- gpt-5.4-2026-03-05high · not in the catalog0.77115.98
- moonshotai/kimi-k2.6unstated0.77110.82
- claude-opus-4-5-20251101unstated · not in the catalog0.7799.92
- qwen3.6-max-previewunstated · not in the catalog0.77106.75
- google/gemini-3.1-pro-preview-customtoolsunstated0.76set aside
- google/gemini-3-flash-previewunstated0.75108.16
- claude-sonnet-4-6unstated · not in the catalog0.75set aside
- openai/gpt-5.3-codexhigh0.75set aside
- z-ai/glm-5.1unstated0.74109.27
- moonshotai/kimi-k2.5unstated0.74set aside
- gpt-5.2-2025-12-11high · not in the catalog0.74110.74
- gpt-5-2025-08-07high · not in the catalog0.74108.61
- claude-opus-4-1-20250805unstated · not in the catalog0.7393.81
- gemini-3-pro-previewunstated · not in the catalog0.73110.74
- z-ai/glm-5unstated0.7299.96
- gpt-5-2025-08-07medium · not in the catalog0.71108.12
- claude-sonnet-4-5-20250929unstated · not in the catalog0.7193.21
- claude-opus-4-20250514unstated · not in the catalog0.7192.48
- gpt-5.1-2025-11-13high · not in the catalog0.68109.94
- gpt-5-mini-2025-08-07medium · not in the catalog0.65103.83
- o3-2025-04-16medium · not in the catalog0.62101.11
- claude-3-7-sonnet-20250219unstated · not in the catalog0.6187.22
- qwen/qwen3.6-plusunstated0.58106.49
- google/gemini-2.5-prounstated0.58103.73
- gpt-4.1-2025-04-14unstated · not in the catalog0.4991.64
- openai/gpt-4o-2024-11-20unstated0.3180.37
GPQA diamondEpoch AI Benchmarking Hub-1.12 · 1.11 · 342 runs
342 runs scored on it, 266 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-6.2-astramax · not in the catalog0.97135.84
- google/gemini-4-hydrogenxhigh0.97125.83
- google/gemini-4-carbonhigh0.97131.68
- google/gemini-4-argonhigh0.96124.90
- gemini-4-neonxhigh · not in the catalog0.96134.54
- claude-fable-5-2-fallbackmax · not in the catalog0.96set aside
- openai/gpt-6-astramax0.96132.41
- claude-sonnet-5-5max · not in the catalog0.96122.13
- google/gemini-3.8-flashhigh0.95114.21
- openai/gpt-6.1-solmax0.95130.00
- google/gemini-3.7-flashhigh0.95116.01
- gpt-5.4-pro-2026-03-05xhigh · not in the catalog0.95117.92
- google/gemini-3.1-pro-previewhigh0.94113.89
- openai/gpt-6-solmax0.94124.00
- openai/gpt-6.1-terramax0.94130.85
- kimi-k3.1-pro-togethermax · not in the catalog0.94118.88
- google/gemini-3.6-flashhigh0.94110.95
- google/gemini-3.1-pro-previewunstated0.94111.66
- claude-opus-5-5-fallbackmax · not in the catalog0.94set aside
- gpt-5.5-pre-releasexhigh · not in the catalog0.94116.86
- x-ai/grok-4.6high0.94114.02
- gpt-5.5-pro-pre-releasexhigh · not in the catalog0.94117.38
- anthropic/claude-opus-5max0.94122.24
- claude-fable-5-2max · not in the catalog0.94132.06
- claude-fable-5-fallbackmax · not in the catalog0.94set aside
- openai/gpt-5.6-solmax0.93123.35
- x-ai/grok-4.5high0.93111.55
- openai/gpt-5.6-terramax0.93119.24
- gpt-5.4-2026-03-05xhigh · not in the catalog0.93116.23
- deepseek/deepseek-v4.1-flash-1017max0.93116.25
- x-ai/grok-4.6xhigh0.93114.39
- moonshotai/kimi-k3max0.93114.75
- doubao-seed-evolvinghigh · not in the catalog0.93114.76
- anthropic/claude-opus-5unstated0.93115.94
- google/gemini-3.5-flashhigh0.93111.12
- x-ai/grok-4.7xhigh0.93113.32
- qwen3.8-maxxhigh · not in the catalog0.93116.39
- gemini-3-pro-previewunstated · not in the catalog0.93110.74
- qwen/qwen3.8-max-0902xhigh0.92113.51
- moonshotai/kimi-k3high0.92112.20
- z-ai/glm-5.2max0.92110.33
- deepseek/deepseek-v4-pro-0813max0.92113.60
- openai/gpt-5.6-lunamax0.92114.95
- gpt-5.2-2025-12-11xhigh · not in the catalog0.91113.39
- deepseek/deepseek-v4-flash-0731max0.91111.14
- claude-opus-4-8max · not in the catalog0.91116.89
- deepseek/deepseek-v4-prohigh0.91108.20
- minimax/minimax-m3unstated0.91104.58
- qwen/qwen3.7-maxunstated0.91111.06
- z-ai/glm-5.3max0.91112.54
- moonshotai/kimi-k2.6unstated0.91110.82
- openai/gpt-5.5low0.91107.86
- claude-opus-5-5max · not in the catalog0.91128.09
- anthropic/claude-sonnet-5xhigh0.91113.50
- claude-opus-4-6-32kunstated · not in the catalog0.91111.03
- openai/gpt-6-lunamax0.90113.99
- claude-opus-4-7xhigh · not in the catalog0.90113.14
- mistral-large-4-0-osshigh · not in the catalog0.90110.78
- z-ai/glm-5.3-flashmax0.90108.45
- openai/gpt-5.6-sollow0.90110.85
- gpt-5.4-2026-03-05high · not in the catalog0.90115.98
- z-ai/glm-5.1unstated0.90109.27
- deepseek/deepseek-v4.1-flashmax0.90116.49
- muse-sparkunstated · not in the catalog0.90110.41
- deepseek/deepseek-v4-promax0.90107.02
- claude-haiku-5-5max · not in the catalog0.90113.88
- doubao-seed-2.1-turbo-260628high · not in the catalog0.90108.84
- google/gemini-3-flash-previewhigh0.89113.70
- doubao-seed-2.1-pro-260628high · not in the catalog0.89109.61
- grok-4.20-0309-reasoningunstated · not in the catalog0.89108.33
- google/gemini-3.5-flashlow0.89110.83
- gpt-5.4-2026-03-05medium · not in the catalog0.89112.04
- x-ai/grok-4.3high0.89107.39
- claude-opus-4-6-64kunstated · not in the catalog0.89111.92
- muse-spark-open-816b-a42bxhigh · not in the catalog0.89113.92
- thinkingmachines/inkling-smallxhigh0.89106.54
- qwen/qwen3.6-plusunstated0.88106.49
- claude-opus-4-6max · not in the catalog0.88111.69
- claude-opus-4-8low · not in the catalog0.88111.54
- thinkingmachines/inklingxhigh0.88107.98
- doubao-seed-2.0-pro-260215high · not in the catalog0.88108.12
- gpt-5.2-2025-12-11high · not in the catalog0.88110.74
- doubao-seed-2.0-lite-260428high · not in the catalog0.88106.42
- mistralai/mistral-large-4-0high0.88110.63
- anthropic/claude-opus-5low0.88109.13
- moonshotai/kimi-k2.7-codeunstated0.88107.73
- qwen/qwen3.7-plusunstated0.88107.80
- gpt-5.2-2025-12-11medium · not in the catalog0.88108.75
- z-ai/glm-5.2low0.88105.42
- z-ai/glm-5unstated0.8899.96
- gpt-5.1-2025-11-13high · not in the catalog0.88109.94
- fireworks/kimi-k2p5unstated · not in the catalog0.88107.32
- openai/gpt-5.6-terralow0.87106.93
- claude-sonnet-4-6-32kunstated · not in the catalog0.87set aside
- qwen3.6-max-previewunstated · not in the catalog0.87106.75
- grok-4-0709unstated · not in the catalog0.87104.16
- gpt-5.4-mini-2026-03-17xhigh · not in the catalog0.87106.91
- google/gemini-3.5-flashminimal0.86104.41
- google/gemini-3.6-flashlow0.86107.80
- qwen/qwen3.5-397b-a17bnone0.86103.55
- claude-opus-4-7max · not in the catalog0.86113.03
- gpt-5-2025-08-07high · not in the catalog0.86108.61
- claude-opus-4-5-20251101-32kunstated · not in the catalog0.86104.05
- anthropic/claude-fable-5max0.86122.58
- google/gemini-3.6-flashminimal0.86105.55
- qwen/qwen3.5-397b-a17bunstated0.86104.40
- qwen/qwen3.6-27bunstated0.86103.72
- claude-opus-4-5-20251101-16kunstated · not in the catalog0.85104.70
- claude-opus-4-8none · not in the catalog0.85104.31
- gpt-5-2025-08-07medium · not in the catalog0.85108.12
- nemotron-3-ultraunstated · not in the catalog0.85105.62
- google/gemini-2.5-prounstated0.85103.73
- gpt-5.1-2025-11-13medium · not in the catalog0.85105.41
- moonshotai/kimi-k3low0.85100.12
- qwen/qwen3.6-27bnone0.85100.32
- qwen/qwen3.6-35b-a3bnone0.85100.22
- gemini-2.5-pro-preview-06-05unstated · not in the catalog0.85103.27
- gpt-5.4-2026-03-05low · not in the catalog0.85104.16
- qwen3.5-plusunstated · not in the catalog0.85105.31
- kimi-k2-thinking-turbounstated · not in the catalog0.84102.46
- qwen/qwen3.6-35b-a3bunstated0.84102.04
- gemini-2.5-pro-exp-03-25unstated · not in the catalog0.84set aside
- gpt-5.4-mini-2026-03-17high · not in the catalog0.84106.09
- qwen/qwen3.5-35b-a3bunstated0.83set aside
- deepseek-reasonerunstated · not in the catalog0.83set aside
- anthropic/claude-fable-5high0.83119.29
- google/gemini-3.5-flash-litehigh0.8399.97
- qwen/qwen3.6-flashunstated0.83100.95
- claude-sonnet-4-6high · not in the catalog0.83102.30
- claude-sonnet-4-6medium · not in the catalog0.83103.01
- z-ai/glm-4.7unstated0.83104.07
- google/gemini-3-flash-previewunstated0.83108.16
- openai/gpt-5.6-solnone0.83100.98
- gpt-5.2-2025-12-11low · not in the catalog0.83104.89
- gpt-5.5-instantunstated · not in the catalog0.8399.89
- openai/gpt-5.6-lunalow0.82101.92
- qwen/qwen3.7-flashunstated0.82102.16
- claude-sonnet-4-5-20250929-59kunstated · not in the catalog0.82101.26
- qwen3.5-flashunstated · not in the catalog0.8299.58
- google/gemini-3.1-flash-litehigh0.82102.13
- qwen/qwen3.7-plusnone0.82103.17
- o3-2025-04-16high · not in the catalog0.82102.96
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.82101.18
- minimax/minimax-m3none0.8190.84
- qwen/qwen3.5-35b-a3bnone0.81set aside
- qwen/qwen3.7-flashnone0.81100.54
- o3-2025-04-16medium · not in the catalog0.81101.11
- claude-opus-4-5-20251101unstated · not in the catalog0.8199.92
- anthropic/claude-sonnet-5max0.80112.43
- qwen/qwen3-235b-a22b-thinking-2507unstated0.80104.10
- o3-2025-04-16low · not in the catalog0.8099.37
- o4-mini-2025-04-16high · not in the catalog0.80103.35
- qwen/qwen3.5-9bunstated0.79set aside
- qwen/qwen3.5-9bnone0.79set aside
- anthropic/claude-fable-5low0.79110.20
- claude-sonnet-4-5-20250929-16kunstated · not in the catalog0.79set aside
- claude-sonnet-4-6max · not in the catalog0.79101.17
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog0.7995.18
- gpt-5.4-nano-2026-03-17high · not in the catalog0.78106.76
- claude-sonnet-4-20250514-32kunstated · not in the catalog0.78set aside
- claude-sonnet-4-20250514-59kunstated · not in the catalog0.7899.17
- o4-mini-2025-04-16medium · not in the catalog0.78101.93
- seed-1.6unstated · not in the catalog0.7897.70
- openai/gpt-5.5none0.7797.11
- openai/gpt-5.6-terranone0.7795.57
- claude-opus-4-1-20250805-16kunstated · not in the catalog0.7796.39
- o3-mini-2025-01-31high · not in the catalog0.77102.45
- claude-3-7-sonnet-20250219-16kunstated · not in the catalog0.7793.20
- claude-3-7-sonnet-20250219-32kunstated · not in the catalog0.7794.40
- claude-opus-4-1-20250805-27kunstated · not in the catalog0.7797.29
- o1-2024-12-17high · not in the catalog0.7798.27
- deepseek/deepseek-r1-0528unstated0.7698.81
- claude-opus-4-20250514-16kunstated · not in the catalog0.76set aside
- grok-3-mini-betalow · not in the catalog0.7694.71
- google/gemini-3.5-flash-litelow0.7698.68
- google/gemma-4-31b-itminimal0.76set aside
- claude-sonnet-4-20250514-16kunstated · not in the catalog0.76set aside
- o1-2024-12-17medium · not in the catalog0.7697.81
- openai/gpt-oss-120bhigh · not in the catalog0.76set aside
- o4-mini-2025-04-16low · not in the catalog0.75100.61
- gpt-5-mini-2025-08-07high · not in the catalog0.75104.57
- gpt-5.4-2026-03-05none · not in the catalog0.7596.15
- grok-3-mini-betahigh · not in the catalog0.7597.03
- o3-mini-2025-01-31medium · not in the catalog0.7497.94
- google/gemini-3.1-flash-litelow0.7494.94
- google/gemini-3.5-flash-liteminimal0.7494.95
- o1-2024-12-17low · not in the catalog0.7494.71
- google/gemini-3.1-flash-liteminimal0.7493.68
- claude-sonnet-4-5-20250929unstated · not in the catalog0.7493.21
- deepseek/deepseek-v4-pronone0.7394.28
- google/gemma-4-26b-a4b-itminimal0.73set aside
- claude-opus-4-1-20250805unstated · not in the catalog0.7393.81
- gpt-5.2-2025-12-11none · not in the catalog0.7396.53
- qwen3-max-2025-09-23unstated · not in the catalog0.73100.03
- gpt-5.4-nano-2026-03-17low · not in the catalog0.7298.40
- gpt-5-2025-08-07minimal · not in the catalog0.7295.14
- gpt-5-mini-2025-08-07minimal · not in the catalog0.7293.86
- gpt-5-mini-2025-08-07medium · not in the catalog0.72103.83
- seed-oss-36b-instructunstated · not in the catalog0.72set aside
- deepseek/deepseek-chatunstated0.71set aside
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.7196.01
- z-ai/glm-5.2none0.7196.22
- qwen3-235b-a22bunstated · not in the catalog0.7187.56
- qwen3-30b-a3b-thinking-2507unstated · not in the catalog0.70set aside
- gpt-5-nano-2025-08-07high · not in the catalog0.69100.45
- deepseek-r1unstated · not in the catalog0.6996.08
- claude-opus-4-20250514unstated · not in the catalog0.6992.48
- gpt-4.5-preview-2025-02-27unstated · not in the catalog0.6990.75
- o3-mini-2025-01-31low · not in the catalog0.6891.91
- deepseek-v3-0324unstated · not in the catalog0.6890.12
- grok-3-betaunstated · not in the catalog0.6894.65
- gpt-5-nano-2025-08-07medium · not in the catalog0.6799.03
- llama-4-maverick-17b-128e-instruct-fp8unstated · not in the catalog0.6787.21
- gpt-4.1-2025-04-14unstated · not in the catalog0.6791.64
- claude-sonnet-4-20250514unstated · not in the catalog0.6790.16
- gemini-2.5-pro-preview-05-06unstated · not in the catalog0.67102.43
- gpt-5.1-2025-11-13none · not in the catalog0.6791.90
- claude-3-7-sonnet-20250219unstated · not in the catalog0.6687.22
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.6693.48
- qwen/qwen3-32bunstated0.6696.21
- gemini-2.0-pro-exp-02-05unstated · not in the catalog0.6690.84
- qwq-plusunstated · not in the catalog0.65set aside
- qwq-32bunstated · not in the catalog0.6593.18
- deepseek-r1-distill-qwen-32bunstated · not in the catalog0.64set aside
- gemini-2.0-flash-001unstated · not in the catalog0.6489.75
- gpt-5.4-mini-2026-03-17none · not in the catalog0.6490.48
- qwen/qwen3-14bunstated0.64set aside
- openai/gpt-5.6-lunanone0.6495.93
- o1-mini-2024-09-12high · not in the catalog0.6292.53
- qwen/qwen3-30b-a3bunstated0.62set aside
- openai/gpt-oss-20bmedium0.61set aside
- z-ai/glm-4.7-flashunstated0.61set aside
- claude-haiku-4-5-20251001unstated · not in the catalog0.6090.99
- mistral-medium-2505unstated · not in the catalog0.6089.41
- o1-mini-2024-09-12medium · not in the catalog0.6091.51
- gpt-5-nano-2025-08-07low · not in the catalog0.5892.74
- gemini-1.5-pro-002unstated · not in the catalog0.5786.80
- gemini-2.0-flash-thinking-exp-01-21unstated · not in the catalog0.5792.14
- qwen3-8bunstated · not in the catalog0.57set aside
- deepseek-v3unstated · not in the catalog0.5784.96
- qwen-max-2025-01-25unstated · not in the catalog0.5685.11
- microsoft/phi-4unstated0.5684.64
- magistral-small-2506unstated · not in the catalog0.5688.63
- deepseek-r1-distill-llama-70bunstated · not in the catalog0.5693.55
- qwen/qwen3-30b-a3b-instruct-2507unstated0.56set aside
- gpt-5.4-nano-2026-03-17none · not in the catalog0.5691.33
- claude-3-5-sonnet-20241022unstated · not in the catalog0.5585.31
- qwen/qwen3-32bnone0.54set aside
- claude-3-5-sonnet-20240620unstated · not in the catalog0.5481.59
- grok-2-1212unstated · not in the catalog0.5483.89
- qwen/qwen3-14bnone0.53set aside
- openai/gpt-oss-20blow0.53set aside
- qwen3-4bunstated · not in the catalog0.52set aside
- llama-4-scout-17b-16e-instructunstated · not in the catalog0.5282.99
- mistral-large-2411unstated · not in the catalog0.5181.24
- llama-3.1-405b-instructunstated · not in the catalog0.5181.49
- qwen/qwen3-30b-a3bnone0.50set aside
- o1-preview-2024-09-12unstated · not in the catalog0.5089.08
- openai/gpt-4o-2024-08-06unstated0.4982.76
- qwen2.5-72b-instructunstated · not in the catalog0.4983.11
- mistral-small-3.2-2506unstated · not in the catalog0.49set aside
- mistralai/mistral-large-2407unstated0.4980.40
- gpt-4.1-nano-2025-04-14unstated · not in the catalog0.4986.22
- openai/gpt-4o-2024-05-13unstated0.4981.01
- gpt-5-nano-2025-08-07minimal · not in the catalog0.4888.08
- qwen-plus-2025-01-25unstated · not in the catalog0.4884.76
- openai/gpt-4o-2024-11-20unstated0.4880.37
- google/gemma-3-27b-itunstated0.4885.05
- magistral-small-2509unstated · not in the catalog0.48set aside
- mistral-small-2503unstated · not in the catalog0.4780.18
- qwen3-8bnone · not in the catalog0.47set aside
- meta-llama/llama-3.3-70b-instructunstated0.4779.22
- gemini-1.5-flash-002unstated · not in the catalog0.4783.86
- mistral-small-3-2501unstated · not in the catalog0.47set aside
- claude-3-opus-20240229unstated · not in the catalog0.4778.88
- gpt-4-turbo-2024-04-09unstated · not in the catalog0.4780.07
- llama-3.1-tulu-3-70b-dpounstated · not in the catalog0.4679.20
- qwen2.5-32b-instructunstated · not in the catalog0.4681.76
- openai/gpt-oss-20bhigh0.46set aside
- gemini-1.5-pro-001unstated · not in the catalog0.4679.25
- qwen3-4b-instruct-2507unstated · not in the catalog0.46set aside
- mistral-small-2501unstated · not in the catalog0.4579.62
- z-ai/glm-4.7-flashnone0.45set aside
- deepseek-r1-distill-qwen-14bunstated · not in the catalog0.4592.30
- meta-llama/llama-3.1-70b-instructunstated0.4477.86
- qwen3-4bnone · not in the catalog0.44set aside
- wizardlm-2-8x22bunstated · not in the catalog0.43set aside
- gpt-4-1106-previewunstated · not in the catalog0.42set aside
- gpt-4-0125-previewunstated · not in the catalog0.42set aside
- mistral-small-3.1-2503unstated · not in the catalog0.42set aside
- qwen-turbo-2024-11-01unstated · not in the catalog0.4281.30
- llama-3.2-90b-vision-instructunstated · not in the catalog0.4177.93
- qwen2-72b-instructunstated · not in the catalog0.41set aside
- claude-3-sonnet-20240229unstated · not in the catalog0.4173.46
- meta-llama-3-70b-instructunstated · not in the catalog0.4174.91
- gemini-1.5-flash-001unstated · not in the catalog0.4075.35
- google/gemma-3-12b-itunstated0.39set aside
- mistral-large-2402unstated · not in the catalog0.3974.62
- claude-3-5-haiku-20241022unstated · not in the catalog0.3880.42
- qwen3-1.7bunstated · not in the catalog0.38set aside
- openai/gpt-4o-mini-2024-07-18unstated0.3880.10
- hermes-2-theta-llama-3-70bunstated · not in the catalog0.3774.21
- google/gemma-2-27b-itunstated0.3675.03
- claude-3-haiku-20240307unstated · not in the catalog0.3671.87
- gpt-4-0314unstated · not in the catalog0.36set aside
- qwen2.5-7b-instructunstated · not in the catalog0.35set aside
- claude-2.0unstated · not in the catalog0.3570.91
- open-mixtral-8x22bunstated · not in the catalog0.34set aside
- gemini-1.0-pro-001unstated · not in the catalog0.3470.18
- eurus-2-7b-primeunstated · not in the catalog0.34set aside
- deepseek-r1-distill-qwen-1.5bunstated · not in the catalog0.34set aside
- claude-2.1unstated · not in the catalog0.33set aside
- gemini-1.5-flash-8b-001unstated · not in the catalog0.33set aside
- dbrx-instructunstated · not in the catalog0.33set aside
- yi-1.5-34b-chatunstated · not in the catalog0.32set aside
- qwen1.5-32b-chatunstated · not in the catalog0.31set aside
- gpt-4-0613unstated · not in the catalog0.3174.56
- qwen3-1.7bnone · not in the catalog0.31set aside
- mixtral-8x7b-instruct-v0.1unstated · not in the catalog0.31set aside
- open-mistral-nemo-2407unstated · not in the catalog0.30set aside
- open-mixtral-8x7bunstated · not in the catalog0.30set aside
- qwen1.5-72b-chatunstated · not in the catalog0.29set aside
- granite-4.0-microunstated · not in the catalog0.28set aside
- gpt-3.5-turbo-1106unstated · not in the catalog0.28set aside
- phi-3-medium-128k-instructunstated · not in the catalog0.28set aside
- gemma-2-9b-itunstated · not in the catalog0.2772.14
- gpt-3.5-turbo-0125unstated · not in the catalog0.2769.92
- ministral-8b-2410unstated · not in the catalog0.27set aside
- meta-llama/llama-3.1-8b-instructunstated0.2772.89
- llama-2-70b-chat-hfunstated · not in the catalog0.2664.40
- meta-llama-3-8b-instructunstated · not in the catalog0.2667.01
- ministral-3b-2410unstated · not in the catalog0.25set aside
- deepseek-llm-67b-chatunstated · not in the catalog0.2566.23
- granite-4.0-1bunstated · not in the catalog0.24set aside
- meta-llama/llama-3.2-1b-instructunstated0.24set aside
- google/gemma-3-4b-itunstated0.23set aside
- gemma-3-1b-itunstated · not in the catalog0.20set aside
- mistral-7b-instruct-v0.3unstated · not in the catalog0.1560.84
- yi-34b-chatunstated · not in the catalog0.15set aside
- open-mistral-7bunstated · not in the catalog0.13set aside
- granite-4.0-350munstated · not in the catalog0.11set aside
- deepseek-r1-0528-qwen3-8bunstated · not in the catalog0.09set aside
MATH level 5Epoch AI Benchmarking Hub-1.25 · 2.51 · 108 runs
108 runs scored on it, 91 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- gpt-5-2025-08-07high · not in the catalog0.98108.61
- gpt-5-2025-08-07medium · not in the catalog0.98108.12
- gpt-5-mini-2025-08-07high · not in the catalog0.98104.57
- o4-mini-2025-04-16high · not in the catalog0.98103.35
- o3-2025-04-16high · not in the catalog0.98102.96
- claude-sonnet-4-5-20250929-32kunstated · not in the catalog0.98101.18
- qwen3-max-2025-09-23unstated · not in the catalog0.97100.03
- gpt-5-mini-2025-08-07medium · not in the catalog0.97103.83
- deepseek/deepseek-r1-0528unstated0.9798.81
- o3-mini-2025-01-31high · not in the catalog0.96102.45
- claude-haiku-4-5-20251001-32kunstated · not in the catalog0.9696.01
- gemini-2.5-pro-preview-05-06unstated · not in the catalog0.96102.43
- gemini-2.5-pro-preview-03-25unstated · not in the catalog0.96set aside
- gpt-5-nano-2025-08-07medium · not in the catalog0.9599.03
- o3-mini-2025-01-31medium · not in the catalog0.9597.94
- gpt-5-nano-2025-08-07high · not in the catalog0.95100.45
- o1-2024-12-17high · not in the catalog0.9598.27
- o1-2024-12-17medium · not in the catalog0.9497.81
- deepseek-r1unstated · not in the catalog0.9396.08
- claude-3-7-sonnet-20250219-64kunstated · not in the catalog0.9195.18
- grok-3-mini-betalow · not in the catalog0.9194.71
- claude-3-7-sonnet-20250219-32kunstated · not in the catalog0.9094.40
- deepseek-r1-distill-llama-70bunstated · not in the catalog0.9093.55
- o1-mini-2024-09-12high · not in the catalog0.8992.53
- grok-3-betaunstated · not in the catalog0.8994.65
- grok-3-mini-betahigh · not in the catalog0.8897.03
- gpt-4.1-mini-2025-04-14unstated · not in the catalog0.8793.48
- deepseek-r1-distill-qwen-14bunstated · not in the catalog0.8792.30
- claude-haiku-4-5-20251001unstated · not in the catalog0.8790.99
- claude-3-7-sonnet-20250219-16kunstated · not in the catalog0.8693.20
- claude-opus-4-20250514unstated · not in the catalog0.8592.48
- claude-sonnet-4-20250514unstated · not in the catalog0.8490.16
- o1-mini-2024-09-12medium · not in the catalog0.8491.51
- gemini-2.0-pro-exp-02-05unstated · not in the catalog0.8390.84
- gpt-4.1-2025-04-14unstated · not in the catalog0.8391.64
- gemini-2.0-flash-001unstated · not in the catalog0.8289.75
- o1-preview-2024-09-12unstated · not in the catalog0.8289.08
- mistral-medium-2505unstated · not in the catalog0.8289.41
- gpt-4.5-preview-2025-02-27unstated · not in the catalog0.7990.75
- deepseek-v3-0324unstated · not in the catalog0.7690.12
- google/gemma-3-27b-itunstated0.7485.05
- llama-4-maverick-17b-128e-instruct-fp8unstated · not in the catalog0.7387.21
- gemini-1.5-pro-002unstated · not in the catalog0.7086.80
- gpt-4.1-nano-2025-04-14unstated · not in the catalog0.7086.22
- qwen3-235b-a22bunstated · not in the catalog0.6987.56
- claude-3-7-sonnet-20250219unstated · not in the catalog0.6887.22
- qwen-max-2025-01-25unstated · not in the catalog0.6785.11
- qwen-plus-2025-01-25unstated · not in the catalog0.6584.76
- microsoft/phi-4unstated0.6584.64
- deepseek-v3unstated · not in the catalog0.6584.96
- grok-2-1212unstated · not in the catalog0.6483.89
- qwen2.5-72b-instructunstated · not in the catalog0.6383.11
- llama-4-scout-17b-16e-instructunstated · not in the catalog0.6282.99
- gemini-1.5-flash-002unstated · not in the catalog0.6283.86
- claude-3-5-sonnet-20241022unstated · not in the catalog0.5785.31
- qwen-turbo-2024-11-01unstated · not in the catalog0.5681.30
- qwen2.5-32b-instructunstated · not in the catalog0.5681.76
- openai/gpt-4o-2024-08-06unstated0.5382.76
- openai/gpt-4o-mini-2024-07-18unstated0.5380.10
- claude-3-5-sonnet-20240620unstated · not in the catalog0.5281.59
- openai/gpt-4o-2024-05-13unstated0.5181.01
- mistral-large-2411unstated · not in the catalog0.5081.24
- openai/gpt-4o-2024-11-20unstated0.5080.37
- llama-3.1-405b-instructunstated · not in the catalog0.5081.49
- mistral-small-2503unstated · not in the catalog0.4780.18
- gpt-4-turbo-2024-04-09unstated · not in the catalog0.4780.07
- claude-3-5-haiku-20241022unstated · not in the catalog0.4680.42
- mistralai/mistral-large-2407unstated0.4580.40
- mistral-small-2501unstated · not in the catalog0.4579.62
- llama-3.1-tulu-3-70b-dpounstated · not in the catalog0.4379.20
- meta-llama/llama-3.3-70b-instructunstated0.4279.22
- gemini-1.5-pro-001unstated · not in the catalog0.4179.25
- gpt-4-1106-previewunstated · not in the catalog0.40set aside
- llama-3.2-90b-vision-instructunstated · not in the catalog0.3977.93
- qwen2-72b-instructunstated · not in the catalog0.39set aside
- claude-3-opus-20240229unstated · not in the catalog0.3778.88
- meta-llama/llama-3.1-70b-instructunstated0.3777.86
- gpt-4-0125-previewunstated · not in the catalog0.35set aside
- google/gemma-2-27b-itunstated0.2875.03
- wizardlm-2-8x22bunstated · not in the catalog0.26set aside
- yi-1.5-34b-chatunstated · not in the catalog0.25set aside
- gemini-1.5-flash-001unstated · not in the catalog0.2575.35
- mistral-large-2402unstated · not in the catalog0.2474.62
- open-mixtral-8x22bunstated · not in the catalog0.24set aside
- gpt-4-0613unstated · not in the catalog0.2374.56
- meta-llama/llama-3.1-8b-instructunstated0.2372.89
- hermes-2-theta-llama-3-70bunstated · not in the catalog0.2374.21
- meta-llama-3-70b-instructunstated · not in the catalog0.2374.91
- gemma-2-9b-itunstated · not in the catalog0.2172.14
- claude-3-sonnet-20240229unstated · not in the catalog0.1873.46
- phi-3-medium-128k-instructunstated · not in the catalog0.18set aside
- gpt-3.5-turbo-1106unstated · not in the catalog0.16set aside
- ministral-8b-2410unstated · not in the catalog0.15set aside
- claude-3-haiku-20240307unstated · not in the catalog0.1571.87
- ministral-3b-2410unstated · not in the catalog0.14set aside
- claude-2.0unstated · not in the catalog0.1270.91
- dbrx-instructunstated · not in the catalog0.12set aside
- gpt-3.5-turbo-0125unstated · not in the catalog0.1269.92
- gemini-1.0-pro-001unstated · not in the catalog0.1170.18
- open-mistral-nemo-2407unstated · not in the catalog0.11set aside
- open-mixtral-8x7bunstated · not in the catalog0.10set aside
- mixtral-8x7b-instruct-v0.1unstated · not in the catalog0.09set aside
- deepseek-llm-67b-chatunstated · not in the catalog0.0666.23
- meta-llama-3-8b-instructunstated · not in the catalog0.0667.01
- yi-34b-chatunstated · not in the catalog0.05set aside
- open-mistral-7bunstated · not in the catalog0.04set aside
- mistral-7b-instruct-v0.3unstated · not in the catalog0.0460.84
- llama-2-70b-chat-hfunstated · not in the catalog0.0364.40
typosLiveBench-1.53 · 0.54 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-fable-5max94122.58
- claude-fable-5-1max · not in the catalog94125.12
- anthropic/claude-opus-5max92122.24
- google/gemini-3.7-flashhigh90116.01
- google/gemini-3.8-flashhigh90114.21
- google/gemini-3.6-flashhigh88110.95
- openai/gpt-5.5xhigh88120.29
- google/gemini-3.5-flashhigh86111.12
- meta/muse-spark-1.3xhigh86116.79
- openai/gpt-5.6-terramax86119.24
- openai/gpt-6.1-solmax86130.00
- claude-sonnet-5-5max · not in the catalog86122.13
- qwen3.8-maxunstated · not in the catalog86118.08
- deepseek/deepseek-v4-flash-vision-expunstated84116.36
- minimax/minimax-m3unstated84104.58
- moonshotai/kimi-k3unstated84116.54
- openai/gpt-5.6-solmax84123.35
- openai/gpt-6.1-solxhigh84124.64
- qwen/qwen3.7-maxunstated84111.06
- claude-opus-4-6-thinking-autohigh · not in the catalog84113.04
- claude-sonnet-5-5xhigh · not in the catalog84119.88
- smaug-miniunstated · not in the catalog84114.73
- x-ai/grok-4.5unstated84114.74
- x-ai/grok-4.6unstated84117.62
- z-ai/glm-5.3unstated84112.92
- deepseek/deepseek-v4-flash-0731unstated82113.47
- deepseek/deepseek-v4-pro-0813unstated82118.60
- deepseek/deepseek-v4.1-flashmax82116.49
- google/gemini-3.1-pro-previewhigh82113.89
- openai/gpt-5.4xhigh82119.02
- openai/gpt-6-astramax82132.41
- qwen/qwen3.6-plusunstated82106.49
- claude-opus-5-5max · not in the catalog82128.09
- gpt-5.2-2025-12-11high · not in the catalog82110.74
- smaug-agenticunstated · not in the catalog82116.30
- deepseek/deepseek-v4-prounstated80111.64
- moonshotai/kimi-k2.7-codeunstated80107.73
- claude-opus-4-7xhigh · not in the catalog80113.14
- claude-opus-5-5xhigh · not in the catalog80125.14
- x-ai/grok-build-0.1unstated80106.25
- nvidia/nemotron-3-ultra-550b-a55bunstated78105.72
- openai/gpt-6-solmax78124.00
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog78110.65
- claude-opus-4-8max · not in the catalog78116.89
- kimi-k2.6-thinkingunstated · not in the catalog78107.26
- muse-wonder-1.0max · not in the catalog78120.86
- qwen3.8-flash-nextunstated · not in the catalog78113.12
- smaug-flashunstated · not in the catalog78116.63
- x-ai/grok-4.3unstated78103.07
- z-ai/glm-5.3-flashunstated78108.42
- anthropic/claude-sonnet-5xhigh76113.50
- qwen/qwen3.8-27bunstated76112.72
- thinkingmachines/inklingxhigh76107.98
- deepseek/deepseek-v4-flashunstated74102.96
- google/gemini-3.5-flash-litehigh7499.97
- meta/muse-spark-1.2xhigh74117.04
- ox-alphamax · not in the catalog74106.05
- z-ai/glm-5.2unstated74109.40
- claude-sonnet-4-6-thinking-automedium · not in the catalog72114.06
- openai/gpt-5.2-codexunstated70110.02
- openai/gpt-5.4-minixhigh70104.22
- openai/gpt-5.6-lunamax70114.95
- qwen/qwen3.6-27bunstated70103.72
- x-ai/grok-4.7xhigh70113.32
- openai/gpt-6-lunamax64113.99
- claude-haiku-5-5max · not in the catalog64113.88
- meta/muse-spark-1.1xhigh62114.21
- mistral-large-4high · not in the catalog60109.81
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog5887.31
- openai/gpt-5.4-nanoxhigh54110.30
- claude-haiku-5-5xhigh · not in the catalog54112.88
code_generationLiveBench-1.62 · 0.50 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-sonnet-5-5max · not in the catalog95.78122.13
- claude-sonnet-5-5xhigh · not in the catalog92.96119.88
- anthropic/claude-fable-5max91.55122.58
- claude-opus-5-5max · not in the catalog91.55128.09
- claude-opus-5-5xhigh · not in the catalog91.55125.14
- claude-fable-5-1max · not in the catalog90.14125.12
- claude-opus-4-7xhigh · not in the catalog85.92113.14
- smaug-agenticunstated · not in the catalog84.51116.30
- anthropic/claude-sonnet-5xhigh83.10113.50
- openai/gpt-5.6-solmax83.10123.35
- openai/gpt-6-solmax83.10124.00
- muse-wonder-1.0max · not in the catalog83.10120.86
- google/gemini-3.7-flashhigh81.69116.01
- meta/muse-spark-1.3xhigh81.69116.79
- openai/gpt-5.5xhigh81.69120.29
- anthropic/claude-opus-5max80.28122.24
- google/gemini-3.5-flashhigh80.28111.12
- moonshotai/kimi-k3unstated80.28116.54
- openai/gpt-5.2-codexunstated80.28110.02
- openai/gpt-6-astramax80.28132.41
- openai/gpt-6.1-solmax80.28130.00
- qwen/qwen3.6-plusunstated80.28106.49
- claude-opus-4-6-thinking-autohigh · not in the catalog80.28113.04
- claude-sonnet-4-6-thinking-automedium · not in the catalog80.28114.06
- smaug-flashunstated · not in the catalog80.28116.63
- openai/gpt-5.6-lunamax78.87114.95
- openai/gpt-6.1-solxhigh78.87124.64
- qwen/qwen3.7-maxunstated78.87111.06
- claude-haiku-5-5xhigh · not in the catalog78.87112.88
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog78.87110.65
- claude-opus-4-8max · not in the catalog78.87116.89
- kimi-k2.6-thinkingunstated · not in the catalog78.87107.26
- z-ai/glm-5.2unstated78.87109.40
- deepseek/deepseek-v4.1-flashmax77.47116.49
- google/gemini-3.6-flashhigh77.47110.95
- openai/gpt-6-lunamax77.47113.99
- qwen/qwen3.8-27bunstated77.47112.72
- claude-haiku-5-5max · not in the catalog77.47113.88
- x-ai/grok-4.6unstated77.47117.62
- z-ai/glm-5.3unstated77.47112.92
- z-ai/glm-5.3-flashunstated77.47108.42
- deepseek/deepseek-v4-flash-0731unstated76.06113.47
- deepseek/deepseek-v4-pro-0813unstated76.06118.60
- google/gemini-3.5-flash-litehigh76.0699.97
- meta/muse-spark-1.1xhigh76.06114.21
- openai/gpt-5.6-terramax76.06119.24
- gpt-5.2-2025-12-11high · not in the catalog76.06110.74
- mistral-large-4high · not in the catalog76.06109.81
- x-ai/grok-4.7xhigh76.06113.32
- google/gemini-3.1-pro-previewhigh74.65113.89
- meta/muse-spark-1.2xhigh74.65117.04
- openai/gpt-5.4xhigh74.65119.02
- thinkingmachines/inklingxhigh74.65107.98
- x-ai/grok-4.3unstated74.65103.07
- openai/gpt-5.4-nanoxhigh74.29110.30
- deepseek/deepseek-v4-flashunstated73.24102.96
- google/gemini-3.8-flashhigh73.24114.21
- ox-alphamax · not in the catalog73.24106.05
- moonshotai/kimi-k2.7-codeunstated71.83107.73
- nvidia/nemotron-3-ultra-550b-a55bunstated71.83105.72
- qwen/qwen3.6-27bunstated71.83103.72
- qwen3.8-maxunstated · not in the catalog71.83118.08
- deepseek/deepseek-v4-prounstated70.42111.64
- smaug-miniunstated · not in the catalog70.42114.73
- deepseek/deepseek-v4-flash-vision-expunstated69.01116.36
- minimax/minimax-m3unstated69.01104.58
- qwen3.8-flash-nextunstated · not in the catalog69.01113.12
- openai/gpt-5.4-minixhigh67.69104.22
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog67.6187.31
- x-ai/grok-4.5unstated67.61114.74
- x-ai/grok-build-0.1unstated63.38106.25
story_generationLiveBench-1.75 · 0.34 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- google/gemini-3.8-flashhigh82.75114.21
- google/gemini-3.7-flashhigh82.25116.01
- qwen3.8-flash-nextunstated · not in the catalog81.37113.12
- claude-opus-4-8max · not in the catalog80.60116.89
- openai/gpt-5.4xhigh79.02119.02
- x-ai/grok-4.6unstated78.27117.62
- claude-fable-5-1max · not in the catalog78.10125.12
- smaug-miniunstated · not in the catalog78114.73
- meta/muse-spark-1.3xhigh77.48116.79
- google/gemini-3.1-pro-previewhigh77.25113.89
- claude-sonnet-5-5xhigh · not in the catalog77.08119.88
- smaug-agenticunstated · not in the catalog76.67116.30
- qwen3.8-maxunstated · not in the catalog76.52118.08
- qwen/qwen3.7-maxunstated76.28111.06
- openai/gpt-5.6-solmax75.92123.35
- google/gemini-3.6-flashhigh75.60110.95
- moonshotai/kimi-k3unstated75.37116.54
- google/gemini-3.5-flashhigh75.20111.12
- deepseek/deepseek-v4-pro-0813unstated74.93118.60
- x-ai/grok-4.7xhigh74.92113.32
- openai/gpt-5.2-codexunstated74.78110.02
- thinkingmachines/inklingxhigh74.18107.98
- muse-wonder-1.0max · not in the catalog74.17120.86
- openai/gpt-6-astramax73.98132.41
- deepseek/deepseek-v4-flash-vision-expunstated73.83116.36
- meta/muse-spark-1.2xhigh73.65117.04
- claude-haiku-5-5xhigh · not in the catalog73.60112.88
- x-ai/grok-4.5unstated73.58114.74
- qwen/qwen3.8-27bunstated73.10112.72
- deepseek/deepseek-v4.1-flashmax72.68116.49
- openai/gpt-5.5xhigh72.55120.29
- x-ai/grok-build-0.1unstated72.53106.25
- z-ai/glm-5.3unstated72.13112.92
- anthropic/claude-fable-5max71.70122.58
- claude-opus-5-5xhigh · not in the catalog71.18125.14
- claude-opus-5-5max · not in the catalog70.92128.09
- ox-alphamax · not in the catalog70.87106.05
- openai/gpt-6.1-solmax70.82130.00
- x-ai/grok-4.3unstated70.60103.07
- openai/gpt-6.1-solxhigh70.57124.64
- deepseek/deepseek-v4-flash-0731unstated70.50113.47
- smaug-flashunstated · not in the catalog69.98116.63
- claude-sonnet-5-5max · not in the catalog69.92122.13
- claude-opus-4-7xhigh · not in the catalog69.40113.14
- nvidia/nemotron-3-ultra-550b-a55bunstated69.23105.72
- meta/muse-spark-1.1xhigh69.15114.21
- openai/gpt-5.6-lunamax69.07114.95
- claude-sonnet-4-6-thinking-automedium · not in the catalog68.90114.06
- moonshotai/kimi-k2.7-codeunstated68.57107.73
- deepseek/deepseek-v4-prounstated67.75111.64
- openai/gpt-5.4-nanoxhigh67.65110.30
- mistral-large-4high · not in the catalog67.50109.81
- google/gemini-3.5-flash-litehigh67.4799.97
- deepseek/deepseek-v4-flashunstated67.27102.96
- openai/gpt-6-solmax66.93124.00
- kimi-k2.6-thinkingunstated · not in the catalog66.88107.26
- openai/gpt-5.6-terramax66.83119.24
- openai/gpt-5.4-minixhigh66.63104.22
- claude-opus-4-6-thinking-autohigh · not in the catalog66.13113.04
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog65.85110.65
- gpt-5.2-2025-12-11high · not in the catalog65.70110.74
- claude-haiku-5-5max · not in the catalog64.53113.88
- minimax/minimax-m3unstated63.55104.58
- openai/gpt-6-lunamax63.28113.99
- z-ai/glm-5.2unstated62.88109.40
- anthropic/claude-sonnet-5xhigh62.32113.50
- anthropic/claude-opus-5max61.32122.24
- qwen/qwen3.6-27bunstated60.13103.72
- z-ai/glm-5.3-flashunstated59.88108.42
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog59.3587.31
- qwen/qwen3.6-plusunstated56.87106.49
olympiadLiveBench-1.88 · 0.76 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-opus-5-5max · not in the catalog93.25128.09
- claude-fable-5-1max · not in the catalog92.97125.12
- anthropic/claude-fable-5max92.84122.58
- anthropic/claude-opus-5max92.80122.24
- claude-sonnet-5-5max · not in the catalog92.47122.13
- openai/gpt-6.1-solmax92.27130.00
- claude-opus-4-8max · not in the catalog92.23116.89
- claude-opus-4-6-thinking-autohigh · not in the catalog92.17113.04
- openai/gpt-6-astramax92.17132.41
- google/gemini-3.8-flashhigh92.16114.21
- claude-opus-5-5xhigh · not in the catalog92.15125.14
- google/gemini-3.1-pro-previewhigh92.10113.89
- google/gemini-3.5-flashhigh91.88111.12
- openai/gpt-6.1-solxhigh91.86124.64
- muse-wonder-1.0max · not in the catalog91.82120.86
- google/gemini-3.7-flashhigh91.79116.01
- meta/muse-spark-1.3xhigh91.74116.79
- claude-sonnet-5-5xhigh · not in the catalog91.72119.88
- openai/gpt-5.6-solmax91.69123.35
- x-ai/grok-4.7xhigh91.65113.32
- moonshotai/kimi-k3unstated91.65116.54
- smaug-agenticunstated · not in the catalog91.56116.30
- openai/gpt-5.6-terramax91.53119.24
- openai/gpt-5.4xhigh91.47119.02
- claude-opus-4-7xhigh · not in the catalog91.38113.14
- openai/gpt-6-solmax91.36124.00
- openai/gpt-5.5xhigh91.35120.29
- deepseek/deepseek-v4-pro-0813unstated91.29118.60
- x-ai/grok-4.6unstated91.21117.62
- qwen3.8-flash-nextunstated · not in the catalog91.17113.12
- qwen3.8-maxunstated · not in the catalog91.15118.08
- smaug-miniunstated · not in the catalog90.85114.73
- anthropic/claude-sonnet-5xhigh90.67113.50
- deepseek/deepseek-v4-prounstated90.62111.64
- google/gemini-3.6-flashhigh90.49110.95
- openai/gpt-6-lunamax90.38113.99
- moonshotai/kimi-k2.7-codeunstated90.30107.73
- smaug-flashunstated · not in the catalog90.20116.63
- deepseek/deepseek-v4-flash-vision-expunstated90.17116.36
- kimi-k2.6-thinkingunstated · not in the catalog90.03107.26
- z-ai/glm-5.3-flashunstated89.84108.42
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog89.46110.65
- mistral-large-4high · not in the catalog89.28109.81
- x-ai/grok-4.5unstated89.24114.74
- deepseek/deepseek-v4-flash-0731unstated89.08113.47
- deepseek/deepseek-v4.1-flashmax89.08116.49
- z-ai/glm-5.2unstated89.05109.40
- openai/gpt-5.6-lunamax88.65114.95
- gpt-5.2-2025-12-11high · not in the catalog88.57110.74
- openai/gpt-5.4-nanoxhigh87.98110.30
- claude-sonnet-4-6-thinking-automedium · not in the catalog87.86114.06
- meta/muse-spark-1.2xhigh87.75117.04
- qwen/qwen3.8-27bunstated87.73112.72
- z-ai/glm-5.3unstated87.51112.92
- meta/muse-spark-1.1xhigh87.50114.21
- openai/gpt-5.2-codexunstated87.00110.02
- claude-haiku-5-5xhigh · not in the catalog86.95112.88
- qwen/qwen3.7-maxunstated86.93111.06
- x-ai/grok-build-0.1unstated86.62106.25
- deepseek/deepseek-v4-flashunstated86.53102.96
- openai/gpt-5.4-minixhigh85.73104.22
- minimax/minimax-m3unstated85.61104.58
- thinkingmachines/inklingxhigh85.40107.98
- claude-haiku-5-5max · not in the catalog83.70113.88
- qwen/qwen3.6-plusunstated82.76106.49
- nvidia/nemotron-3-ultra-550b-a55bunstated82.58105.72
- qwen/qwen3.6-27bunstated82.32103.72
- x-ai/grok-4.3unstated82.26103.07
- ox-alphamax · not in the catalog81.95106.05
- google/gemini-3.5-flash-litehigh81.8099.97
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog58.9287.31
spatialLiveBench-3.10 · 0.85 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-opus-5max100122.24
- anthropic/claude-sonnet-5xhigh100113.50
- google/gemini-3.6-flashhigh100110.95
- google/gemini-3.7-flashhigh100116.01
- google/gemini-3.8-flashhigh100114.21
- meta/muse-spark-1.1xhigh100114.21
- meta/muse-spark-1.3xhigh100116.79
- moonshotai/kimi-k3unstated100116.54
- openai/gpt-5.6-solmax100123.35
- openai/gpt-5.6-terramax100119.24
- qwen/qwen3.6-27bunstated100103.72
- claude-fable-5-1max · not in the catalog100125.12
- claude-opus-4-7xhigh · not in the catalog100113.14
- claude-sonnet-4-6-thinking-automedium · not in the catalog100114.06
- qwen3.8-flash-nextunstated · not in the catalog100113.12
- qwen3.8-maxunstated · not in the catalog100118.08
- smaug-agenticunstated · not in the catalog100116.30
- x-ai/grok-4.5unstated100114.74
- deepseek/deepseek-v4.1-flashmax98116.49
- google/gemini-3.1-pro-previewhigh98113.89
- nvidia/nemotron-3-ultra-550b-a55bunstated98105.72
- openai/gpt-5.4xhigh98119.02
- openai/gpt-5.5xhigh98120.29
- openai/gpt-6-astramax98132.41
- openai/gpt-6.1-solmax98130.00
- openai/gpt-6.1-solxhigh98124.64
- qwen/qwen3.6-plusunstated98106.49
- claude-opus-4-6-thinking-autohigh · not in the catalog98113.04
- claude-opus-4-8max · not in the catalog98116.89
- claude-opus-5-5max · not in the catalog98128.09
- claude-opus-5-5xhigh · not in the catalog98125.14
- claude-sonnet-5-5max · not in the catalog98122.13
- claude-sonnet-5-5xhigh · not in the catalog98119.88
- smaug-flashunstated · not in the catalog98116.63
- smaug-miniunstated · not in the catalog98114.73
- x-ai/grok-4.3unstated98103.07
- x-ai/grok-4.6unstated98117.62
- x-ai/grok-build-0.1unstated98106.25
- anthropic/claude-fable-5max96122.58
- deepseek/deepseek-v4-flashunstated96102.96
- deepseek/deepseek-v4-flash-vision-expunstated96116.36
- deepseek/deepseek-v4-pro-0813unstated96118.60
- google/gemini-3.5-flashhigh96111.12
- meta/muse-spark-1.2xhigh96117.04
- openai/gpt-5.6-lunamax96114.95
- openai/gpt-6-lunamax96113.99
- openai/gpt-6-solmax96124.00
- qwen/qwen3.7-maxunstated96111.06
- qwen/qwen3.8-27bunstated96112.72
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog96110.65
- gpt-5.2-2025-12-11high · not in the catalog96110.74
- muse-wonder-1.0max · not in the catalog96120.86
- ox-alphamax · not in the catalog96106.05
- z-ai/glm-5.3unstated96112.92
- z-ai/glm-5.3-flashunstated96108.42
- deepseek/deepseek-v4-prounstated94111.64
- minimax/minimax-m3unstated94104.58
- openai/gpt-5.2-codexunstated94110.02
- thinkingmachines/inklingxhigh94107.98
- claude-haiku-5-5xhigh · not in the catalog94112.88
- kimi-k2.6-thinkingunstated · not in the catalog94107.26
- mistral-large-4high · not in the catalog94109.81
- x-ai/grok-4.7xhigh94113.32
- z-ai/glm-5.2unstated94109.40
- moonshotai/kimi-k2.7-codeunstated92107.73
- claude-haiku-5-5max · not in the catalog92113.88
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog9287.31
- deepseek/deepseek-v4-flash-0731unstated90113.47
- google/gemini-3.5-flash-litehigh9099.97
- openai/gpt-5.4-minixhigh88104.22
- openai/gpt-5.4-nanoxhigh82110.30
AMPS_HardLiveBench-3.53 · 0.74 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- anthropic/claude-opus-5max99.01122.24
- anthropic/claude-fable-5max99122.58
- google/gemini-3.8-flashhigh99114.21
- meta/muse-spark-1.3xhigh99116.79
- thinkingmachines/inklingxhigh99107.98
- claude-fable-5-1max · not in the catalog99125.12
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog99110.65
- claude-opus-5-5xhigh · not in the catalog99125.14
- muse-wonder-1.0max · not in the catalog99120.86
- qwen3.8-flash-nextunstated · not in the catalog99113.12
- x-ai/grok-4.5unstated99114.74
- z-ai/glm-5.3unstated99112.92
- anthropic/claude-sonnet-5xhigh98113.50
- deepseek/deepseek-v4-flashunstated98102.96
- deepseek/deepseek-v4-prounstated98111.64
- deepseek/deepseek-v4-pro-0813unstated98118.60
- deepseek/deepseek-v4.1-flashmax98116.49
- google/gemini-3.1-pro-previewhigh98113.89
- google/gemini-3.5-flashhigh98111.12
- google/gemini-3.6-flashhigh98110.95
- google/gemini-3.7-flashhigh98116.01
- openai/gpt-5.2-codexunstated98110.02
- openai/gpt-5.4xhigh98119.02
- openai/gpt-5.4-nanoxhigh98110.30
- openai/gpt-5.6-lunamax98114.95
- openai/gpt-5.6-solmax98123.35
- openai/gpt-5.6-terramax98119.24
- openai/gpt-6-astramax98132.41
- openai/gpt-6-lunamax98113.99
- openai/gpt-6-solmax98124.00
- openai/gpt-6.1-solmax98130.00
- openai/gpt-6.1-solxhigh98124.64
- qwen/qwen3.7-maxunstated98111.06
- qwen/qwen3.8-27bunstated98112.72
- claude-haiku-5-5xhigh · not in the catalog98112.88
- claude-opus-4-7xhigh · not in the catalog98113.14
- claude-opus-4-8max · not in the catalog98116.89
- claude-opus-5-5max · not in the catalog98128.09
- claude-sonnet-5-5max · not in the catalog98122.13
- claude-sonnet-5-5xhigh · not in the catalog98119.88
- mistral-large-4high · not in the catalog98109.81
- ox-alphamax · not in the catalog98106.05
- qwen3.8-maxunstated · not in the catalog98118.08
- smaug-agenticunstated · not in the catalog98116.30
- smaug-miniunstated · not in the catalog98114.73
- x-ai/grok-4.7xhigh98113.32
- z-ai/glm-5.2unstated98109.40
- deepseek/deepseek-v4-flash-0731unstated97113.47
- moonshotai/kimi-k2.7-codeunstated97107.73
- moonshotai/kimi-k3unstated97116.54
- openai/gpt-5.4-minixhigh97104.22
- openai/gpt-5.5xhigh97120.29
- qwen/qwen3.6-plusunstated97106.49
- claude-haiku-5-5max · not in the catalog97113.88
- claude-opus-4-6-thinking-autohigh · not in the catalog97113.04
- gpt-5.2-2025-12-11high · not in the catalog97110.74
- kimi-k2.6-thinkingunstated · not in the catalog97107.26
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog9787.31
- smaug-flashunstated · not in the catalog97116.63
- x-ai/grok-4.3unstated97103.07
- x-ai/grok-4.6unstated97117.62
- z-ai/glm-5.3-flashunstated97108.42
- nvidia/nemotron-3-ultra-550b-a55bunstated96105.72
- google/gemini-3.5-flash-litehigh9599.97
- qwen/qwen3.6-27bunstated93103.72
- meta/muse-spark-1.2xhigh89117.04
- deepseek/deepseek-v4-flash-vision-expunstated88116.36
- meta/muse-spark-1.1xhigh82114.21
- claude-sonnet-4-6-thinking-automedium · not in the catalog76114.06
- minimax/minimax-m3unstated75104.58
- x-ai/grok-build-0.1unstated66106.25
math_compLiveBench-3.85 · 0.65 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- claude-haiku-5-5max · not in the catalog98.04113.88
- claude-opus-4-7xhigh · not in the catalog98.04113.14
- claude-opus-4-8max · not in the catalog98.04116.89
- deepseek/deepseek-v4-flashunstated97.06102.96
- deepseek/deepseek-v4-flash-vision-expunstated97.06116.36
- deepseek/deepseek-v4-pro-0813unstated97.06118.60
- meta/muse-spark-1.2xhigh97.06117.04
- meta/muse-spark-1.3xhigh97.06116.79
- openai/gpt-6-astramax97.06132.41
- openai/gpt-6-solmax97.06124.00
- openai/gpt-6.1-solmax97.06130.00
- openai/gpt-6.1-solxhigh97.06124.64
- qwen/qwen3.7-maxunstated97.06111.06
- thinkingmachines/inklingxhigh97.06107.98
- claude-fable-5-1max · not in the catalog97.06125.12
- claude-haiku-5-5xhigh · not in the catalog97.06112.88
- claude-opus-5-5max · not in the catalog97.06128.09
- claude-opus-5-5xhigh · not in the catalog97.06125.14
- claude-sonnet-5-5max · not in the catalog97.06122.13
- claude-sonnet-5-5xhigh · not in the catalog97.06119.88
- mistral-large-4high · not in the catalog97.06109.81
- muse-wonder-1.0max · not in the catalog97.06120.86
- smaug-flashunstated · not in the catalog97.06116.63
- x-ai/grok-4.5unstated97.06114.74
- x-ai/grok-4.6unstated97.06117.62
- x-ai/grok-4.7xhigh97.06113.32
- deepseek/deepseek-v4-flash-0731unstated96.08113.47
- deepseek/deepseek-v4-prounstated96.08111.64
- deepseek/deepseek-v4.1-flashmax96.08116.49
- google/gemini-3.1-pro-previewhigh96.08113.89
- google/gemini-3.7-flashhigh96.08116.01
- meta/muse-spark-1.1xhigh96.08114.21
- nvidia/nemotron-3-ultra-550b-a55bunstated96.08105.72
- openai/gpt-5.5xhigh96.08120.29
- kimi-k2.6-thinkingunstated · not in the catalog96.08107.26
- z-ai/glm-5.2unstated96.08109.40
- z-ai/glm-5.3unstated96.08112.92
- anthropic/claude-fable-5max95.10122.58
- anthropic/claude-sonnet-5xhigh95.10113.50
- google/gemini-3.5-flashhigh95.10111.12
- google/gemini-3.8-flashhigh95.10114.21
- moonshotai/kimi-k3unstated95.10116.54
- openai/gpt-5.2-codexunstated95.10110.02
- openai/gpt-5.4-nanoxhigh95.10110.30
- openai/gpt-5.6-solmax95.10123.35
- openai/gpt-5.6-terramax95.10119.24
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog95.10110.65
- claude-opus-4-6-thinking-autohigh · not in the catalog95.10113.04
- gpt-5.2-2025-12-11high · not in the catalog95.10110.74
- qwen3.8-maxunstated · not in the catalog95.10118.08
- x-ai/grok-4.3unstated95.10103.07
- x-ai/grok-build-0.1unstated95.10106.25
- anthropic/claude-opus-5max94.12122.24
- google/gemini-3.6-flashhigh94.12110.95
- moonshotai/kimi-k2.7-codeunstated94.12107.73
- openai/gpt-5.4xhigh94.12119.02
- openai/gpt-5.4-minixhigh94.12104.22
- openai/gpt-6-lunamax94.12113.99
- qwen/qwen3.8-27bunstated94.12112.72
- claude-sonnet-4-6-thinking-automedium · not in the catalog94.12114.06
- qwen3.8-flash-nextunstated · not in the catalog94.12113.12
- smaug-agenticunstated · not in the catalog94.12116.30
- smaug-miniunstated · not in the catalog94.12114.73
- qwen/qwen3.6-plusunstated93.14106.49
- z-ai/glm-5.3-flashunstated93.14108.42
- google/gemini-3.5-flash-litehigh92.1699.97
- openai/gpt-5.6-lunamax92.16114.95
- qwen/qwen3.6-27bunstated92.16103.72
- minimax/minimax-m3unstated91.18104.58
- ox-alphamax · not in the catalog91.18106.05
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog86.2887.31
tablereformatLiveBench-3.95 · 0.82 · 71 runs
71 runs scored on it, 71 of them ranked, best score first, in the publisher's own unit. A ranked run is a door to its row.
- deepseek/deepseek-v4-flash-0731unstated100113.47
- deepseek/deepseek-v4-prounstated100111.64
- deepseek/deepseek-v4.1-flashmax100116.49
- meta/muse-spark-1.1xhigh100114.21
- meta/muse-spark-1.2xhigh100117.04
- meta/muse-spark-1.3xhigh100116.79
- openai/gpt-5.2-codexunstated100110.02
- openai/gpt-5.4xhigh100119.02
- openai/gpt-5.4-minixhigh100104.22
- openai/gpt-5.5xhigh100120.29
- openai/gpt-5.6-lunamax100114.95
- openai/gpt-5.6-solmax100123.35
- openai/gpt-5.6-terramax100119.24
- openai/gpt-6-astramax100132.41
- openai/gpt-6-lunamax100113.99
- openai/gpt-6-solmax100124.00
- openai/gpt-6.1-solmax100130.00
- openai/gpt-6.1-solxhigh100124.64
- thinkingmachines/inklingxhigh100107.98
- gpt-5.2-2025-12-11high · not in the catalog100110.74
- mistral-large-4high · not in the catalog100109.81
- muse-wonder-1.0max · not in the catalog100120.86
- ox-alphamax · not in the catalog100106.05
- x-ai/grok-4.5unstated100114.74
- x-ai/grok-4.6unstated100117.62
- x-ai/grok-4.7xhigh100113.32
- x-ai/grok-build-0.1unstated100106.25
- anthropic/claude-sonnet-5xhigh98.04113.50
- deepseek/deepseek-v4-flashunstated98.04102.96
- deepseek/deepseek-v4-flash-vision-expunstated98.04116.36
- deepseek/deepseek-v4-pro-0813unstated98.04118.60
- google/gemini-3.1-pro-previewhigh98.04113.89
- google/gemini-3.5-flash-litehigh98.0499.97
- google/gemini-3.6-flashhigh98.04110.95
- google/gemini-3.7-flashhigh98.04116.01
- minimax/minimax-m3unstated98.04104.58
- moonshotai/kimi-k3unstated98.04116.54
- nvidia/nemotron-3-ultra-550b-a55bunstated98.04105.72
- openai/gpt-5.4-nanoxhigh98.04110.30
- qwen/qwen3.6-27bunstated98.04103.72
- qwen/qwen3.6-plusunstated98.04106.49
- qwen/qwen3.7-maxunstated98.04111.06
- claude-haiku-5-5max · not in the catalog98.04113.88
- claude-haiku-5-5xhigh · not in the catalog98.04112.88
- claude-opus-4-5-20251101-thinking-64khigh · not in the catalog98.04110.65
- claude-opus-4-6-thinking-autohigh · not in the catalog98.04113.04
- claude-opus-4-7xhigh · not in the catalog98.04113.14
- claude-opus-5-5max · not in the catalog98.04128.09
- claude-opus-5-5xhigh · not in the catalog98.04125.14
- claude-sonnet-5-5max · not in the catalog98.04122.13
- claude-sonnet-5-5xhigh · not in the catalog98.04119.88
- kimi-k2.6-thinkingunstated · not in the catalog98.04107.26
- qwen3.8-maxunstated · not in the catalog98.04118.08
- smaug-agenticunstated · not in the catalog98.04116.30
- smaug-flashunstated · not in the catalog98.04116.63
- x-ai/grok-4.3unstated98.04103.07
- google/gemini-3.5-flashhigh96.08111.12
- google/gemini-3.8-flashhigh96.08114.21
- qwen/qwen3.8-27bunstated96.08112.72
- claude-sonnet-4-6-thinking-automedium · not in the catalog96.08114.06
- qwen3.8-flash-nextunstated · not in the catalog96.08113.12
- smaug-miniunstated · not in the catalog96.08114.73
- z-ai/glm-5.2unstated96.08109.40
- z-ai/glm-5.3unstated96.08112.92
- z-ai/glm-5.3-flashunstated96.08108.42
- anthropic/claude-fable-5max94.12122.58
- anthropic/claude-opus-5max94.12122.24
- claude-fable-5-1max · not in the catalog94.12125.12
- claude-opus-4-8max · not in the catalog94.12116.89
- moonshotai/kimi-k2.7-codeunstated92.16107.73
- nemotron-3.5-lightning-30b-a3bunstated · not in the catalog86.2887.31
46 columns, hardest first; the words in the head reorder them. A thick span fits inside one standard deviation, and 12 do. A span wider than the shading sorts nothing among the ranked runs, and 5 are wider than the whole ranked range. An end without a cap ran past the axis. Difficulty and slope in the right column are on the fit standardised scale, where zero is the mean run; the marker is the same difficulty on the presented scale; runs is how many the column scored, which is the evidence behind its difficulty. A row opens onto every run scored on the column. A column is named board first because two publishers can name a benchmark the same thing and mean two different sets of questions. The boards themselves, with the licence and the metric each publisher states, are on the benchmarks page.
Each board on this axis is placed against the others through the runs it shares with them: Epoch AI Benchmarking Hub through 56, LiveBench through 37 and Aider polyglot through 19. A group of columns goes on the axis only where 2 runs or more were scored both in it and on the rest, because one shared run fixes where the group sits and not how far apart its runs are. HELM (GPQA - COT correct, IFEval - IFEval Strict Acc, MMLU-Pro - COT correct, Mean score, Omni-MATH - Acc, WildBench - WB Score) and Vectara hallucination leaderboard (factual consistency rate) share 1 run with the rest, zai-org/glm-4.6, and are excluded with the 161 runs scored only there. MTEB (SciFact) shares no run with the rest and is excluded with the 267 runs scored only there.
Over the 64 models on both scales, the two orderings agree at a rank correlation of 0.963, and 8 of them hold the same place on both.
- 01GPT-5.4 Nano40 here · 54 on the index · 14 places apart
- 02Muse Spark 1.218 here · 31 on the index · 13 places apart
- 03DeepSeek V4 Pro 081316 here · 28 on the index · 12 places apart
- 04DeepSeek-V4-Pro33 here · 44 on the index · 11 places apart
- 05Gemini 3.8 Flash26 here · 16 on the index · 10 places apart
A model here is its strongest ranked run, so a run under the 3 column floor is not compared, and a place is counted among the 64 models on both scales rather than among everything either one ranks. Only the order compares: the two fits share a form and nothing else, so their numbers were never going to line up. Each numbered model is a door to its row in the ranked ledger. Computed at render from two committed documents, this fit and the index as read , and the same figure scripts/capability-index.mjs --against-eci prints. The index itself, with the interval Epoch publishes on every figure, is on its own board.
Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.
All 64 models on both scales
- GPT-6.2 Astra1 here1 on the index
- Gemini 4 Carbon3 here2 on the index
- GPT-6 Astra2 here3 on the index
- GPT-6.1 Sol5 here4 on the index
- GPT-6.1 Terra4 here5 on the index
- Gemini 4 Hydrogen6 here6 on the index
- Gemini 4 Argon7 here7 on the index
- Claude Opus 512 here8 on the index
- Claude Fable 511 here9 on the index
- GPT-6 Sol8 here10 on the index
- GPT-5.6 Sol9 here11 on the index
- GPT-5.5 Pro10 here12 on the index
- GPT-5.6 Terra14 here13 on the index
- GPT-5.513 here14 on the index
- DeepSeek V4.1 Flash 101722 here15 on the index
- Gemini 3.8 Flash26 here16 on the index
- Kimi K320 here17 on the index
- Gemini 3.7 Flash23 here18 on the index
- Muse Spark 1.319 here19 on the index
- GPT-5.415 here20 on the index
- Grok 4.617 here21 on the index
- GLM-5.332 here22 on the index
- DeepSeek V4.1 Flash21 here23 on the index
- Claude Sonnet 529 here24 on the index
- GPT-5.6 Luna24 here25 on the index
- GPT-6 Luna28 here26 on the index
- Grok 4.731 here27 on the index
- DeepSeek V4 Pro 081316 here28 on the index
- Gemini 3.5 Flash34 here29 on the index
- Gemini 3.6 Flash36 here30 on the index
- Muse Spark 1.218 here31 on the index
- Grok 4.525 here32 on the index
- Muse Spark 1.126 here33 on the index
- Qwen3.7-Max35 here34 on the index
- Mistral Large 441 here34 on the index
- DeepSeek V4 Flash 073130 here36 on the index
- GLM-5.3-Flash43 here37 on the index
- Muse Spark38 here38 on the index
- GLM-5.239 here39 on the index
- Kimi K2.637 here40 on the index
- Kimi K2.7 Code46 here41 on the index
- Inkling-Small47 here42 on the index
- GLM-5.142 here43 on the index
- DeepSeek-V4-Pro33 here44 on the index
- GPT-5.4 Mini52 here45 on the index
- Inkling44 here46 on the index
- Qwen3.5 397B-A17B51 here47 on the index
- Qwen3.7-Plus45 here48 on the index
- MiniMax-M350 here49 on the index
- Nemotron 3 Ultra48 here50 on the index
- DeepSeek-V4-Flash55 here51 on the index
- Qwen3.6 27B54 here52 on the index
- GLM-558 here53 on the index
- GPT-5.4 Nano40 here54 on the index
- Gemini 3.5 Flash-Lite57 here55 on the index
- Kimi K2 Thinking49 here56 on the index
- Gemini 3.1 Flash-Lite56 here57 on the index
- GLM-4.753 here58 on the index
- DeepSeek-R160 here59 on the index
- Qwen3-235B-A22B62 here60 on the index
- Qwen3-32B59 here61 on the index
- QwQ-32B61 here62 on the index
- DeepSeek-V363 here63 on the index
- Phi-464 here64 on the index
Somebody else's measurements, on a scale fit here.
Fit by Vitna over boards published by others. Every score behind it was produced by somebody else, under the licences named in app/data/boards.json, and Vitna ran none of these evaluations.
Mean 100, 15 to a standard deviation, over the runs in this document. This fit's own scale: not an Epoch Capabilities Index score, and not comparable to one. The orderings are.
The form is Epoch AI's, published in their MIT package and their paper: one capability per run, one difficulty and one slope per benchmark column, fit by maximum likelihood. The ordering this produces is checked against their own Capabilities Index over the models on both scales, and the check is the fourth exhibit above, computed from the committed documents rather than asserted.
Vitna's own question, of what Vitna has measured, still answers none. Every cell behind this scale was produced by somebody else under a licence that permits republishing it, and moving from a publisher's ordering to one fit here changes whose arithmetic it is, not whose work.
Fit over the mirror read , on document shape 2. It is re-derived from the committed boards on every mirror run and rewritten only when the ordering, the set of runs or the evidence behind one of them moves, so an unchanged day leaves this page exactly as it was.
