Get started

GLM 5.3 (batch)

Sold by 47 sellers, timed at 96.4 tok/s, scored 31.7 on the Intelligence Index, on 12 Design Arena boards, and with results on 2 openly licensed boards.

retrieved sha256 · 0bb53a77713f

Models/Z.ai/GLM 5.3 (batch)USD per 1M tokens · as OpenRouter routes it
z-ai/glm-5.3:batchZ.ai · batch variant · Other tokenizer
OpenRouter list price · the routed rate, as the catalog carried itretrieved 17 Nov 16:29 UTC from openrouter.ai ↗
Input$0.281per 1M tokens
Output$1.25per 1M tokens
Cached input$0.062per 1M tokens
Context1.05M1,048,576 tokens
What the list says · the fields OpenRouter publishes for this id12 fields
AuthorZ.ai
Listed
Canonical idz-ai/glm-5.3-20260816
Context1,048,576 tokens
Max output943,718 tokens
Modalitiestext → text
TokenizerOther
Reasoningalways on · effort max, high, low, max by default
Moderationnot moderated by OpenRouter
Variantbatch
Other listingbase listing
What the weights say · read from the model's own repository on Hugging Faceretrieved from huggingface.co ↗
Licenceglm-5.3, tagged other
Parameters753B · 753,329,940,480 declared
Model classGlmMoeDsaForCausalLM
Downloadnot gated
Published

What sellers charge

47 sellers · 30.04x from cheapest to dearest

One tick per seller, on a log scale of input price per million from $0.0699 at Wafer to $2.10 at Parasail. 6 sellers undercut the $0.281 listed; 41 are dearer. A faint tick declares no number format. Every seller is named on the endpoint ledger below.

The same sellers grouped by the number format each declares, on the same scale. The fp4 group alone spans 100% of the range, so the cheap end is not cheap for its number format. A declaration is checked by nobody: this groups the rows and certifies none.

Measured speed

230,032 requests · 46 endpoints · 30-minute window
Throughput96.4 tok/sMedian output rate, weighted across the 46 endpoints serving it by the requests each answered.the speed board
First token513msMedian wait before the first token, over the same window and weighted the same way.how Vitna reads a figure
Endpoint spread40 to 178The slowest and the fastest route serving this model, in tokens per second. Speed is a property of the route, not of the weights.the board, fastest first

OpenRouter's own timing of its routed traffic, weighted by requests, over one 30-minute window read . Every route is on the endpoint ledger below.

Every endpoint

47 priced · 46 timed · cheapest first
  1. Waferwafer/us$0.0699$1.40$0.0689fp4944knot matchednonenonenot matched99.93%
  2. Relacerelace/fp4$0.0703$4.3843$0.0703fp4131k83 tok/s49740ms1,440 req100%
  3. InferenceNetinference-net/fp4$0.084$3.60$0.079fp4944k81 tok/s43407ms9,350 req100%
  4. InferenceNetinference-net/us$0.084$3.60$0.079fp4944k73 tok/s39429ms3,824 req100%
  5. Waferwafer/fp4$0.179$1.40$0.178fp4944knot matchednonenonenot matched100%
  6. AkashMLakashml/fp4$0.19$4.40$0.19fp4131k99 tok/s391.16s541 req100%
  7. DeepInfradeepinfra/fp4/flex · 38% off$0.45$2.00$0.10fp4131knot matchednonenonenot matched100%
  8. Coherecohere/nvfp4$0.50$3.50$0.10nvfp4131k91 tok/s59517ms5,025 req100%
  9. Decartdecart/fp4$0.56$1.81$0.10fp4944k101 tok/s68658ms14,192 req100%
  10. DeepInfradeepinfra/fp4 · 38% off$0.5625$2.50$0.125fp4131knot matchednonenonenot matched99.08%
  11. Tenstorrenttenstorrent/fp8$0.60$2.50$0.10fp8131k135 tok/s321.06s11,887 req100%
  12. Novitanovita/fp8 · 55% off$0.63$1.98$0.117fp8131k60 tok/s371.46s4,128 req100%
  13. Cloudflarecloudflare$0.70$2.20$0.13not declared944k66 tok/s153.3s1,229 req100%
  14. Doubleworddoubleword/nvfp4$0.70$2.20$0.14nvfp4944k53 tok/s35757ms64 req100%
  15. Fireworksfireworks/spot$0.70$2.20$0.13not declared944knot matchednonenonenot matched96.55%
  16. SiliconFlowsiliconflow/fp8$0.70$1.90$0.13fp8262k80 tok/s501.4s137 req100%
  17. CoreWeavecoreweave/nvfp4$0.75$3.00$0.19nvfp4944k91 tok/s50792ms17,299 req99.79%
  18. Sail Researchsail-research/us$0.75$3.00$0.12fp8944knot matchednonenonenot matched100%
  19. DigitalOceandigitalocean$0.91$2.86$0.169not declared128k105 tok/s74452ms4,651 req100%
  20. Featherlessfeatherless · 30% off$0.98$3.08$0.182not declared33k53 tok/s29735ms39 req92.30%
  21. Morphmorph/fp8$1.012$4.723$0.205fp8944k88 tok/s57861ms379 req99.92%
  22. Inceptroninceptron/fp4$1.05$3.50$0.1171fp4944k107 tok/s46625ms2,399 req99.93%
  23. Sail Researchsail-research/fp8$1.05$4.20$0.168fp8944knot matchednonenonenot matched100%
  24. Alibabaalibaba$1.19$3.74$0.238not declared131knot matchednonenonenot matched100%
  25. Friendlifriendli · 10% off$1.26$3.96$0.234not declared944k92 tok/s67303ms2,763 req100%
  26. Baidubaidu/fp8$1.40$4.40$0.26fp8131knot timednonenonenoneno reading
  27. BaseTenbaseten/fp4$1.40$4.40$0.14fp4262knot matchednonenonenot matched100%
  28. BaseTenbaseten/fp4/us$1.40$4.40$0.14fp4262knot matchednonenonenot matched100%
  29. Crusoecrusoe/fp4$1.40$4.40$0.26fp4944k72 tok/s46612ms191 req100%
  30. Fireworksfireworks$1.40$4.40$0.26not declared944knot matchednonenonenot matched100%
  31. Mistralmistral/zdr$1.40$4.40$0.14nvfp4131knot matchednonenonenot matched99.72%
  32. Modalmodal$1.40$4.40$0.26not declared944k99 tok/s55445ms245 req99.82%
  33. Nebiusnebius/fp4$1.40$4.40$0.26fp4944k112 tok/s52374ms865 req100%
  34. Parasailparasail/fp4$1.40$4.40$0.26fp4944knot matchednonenonenot matched99.72%
  35. SambaNovasambanova · 50% off$1.40$4.40$0.07not declared944k48 tok/s201.29s73 req100%
  36. Togethertogether$1.40$4.40$0.26not declared944k103 tok/s70344ms4,510 req99.79%
  37. Venicevenice · 20% off$1.40$4.40$0.26not declared131k104 tok/s44593ms318 req100%
  38. Z.AIz-ai/fp8$1.40$4.40$0.26fp8131k49 tok/s261.8s6,355 req100%
  39. Makoramakora/fp4$1.50$4.40$0.26fp4128k133 tok/s89238ms2,675 req98.61%
  40. Mistralmistral/nvfp4/eu$1.54$4.84$0.154nvfp4131knot matchednonenonenot matched98.91%
  41. Alibabaalibaba/prime$1.68$5.28$0.336not declared131knot matchednonenonenot matched100%
  42. Amazon Bedrockamazon-bedrock/us-east-1$1.68$5.28$0.312not declared131k146 tok/s431.6s60 req99.50%
  43. BaseTenbaseten/fast$2.10$6.60$0.21fp8262knot matchednonenonenot matched100%
  44. BaseTenbaseten/us/fast$2.10$6.60$0.21fp8262knot matchednonenonenot matched100%
  45. Fireworksfireworks/fast$2.10$6.60$0.39not declared944knot matchednonenonenot matched99.88%
  46. ModelRunmodelrun/fp4$2.10$6.60$0.40fp4524k137 tok/s68143ms2,830 req100%
  47. Parasailparasail/fast$2.10$6.60$0.26fp4944knot matchednonenonenot matched98.93%
  48. Alibabanot matchednot matchednot matchednot matchedno offer118 tok/s43.21.49s182 reqno offer
  49. Alibabanot matchednot matchednot matchednot matchedno offer40 tok/s22938ms4,084 reqno offer
  50. BaseTennot matchednot matchednot matchednot matchedno offer144 tok/s81454ms803 reqno offer
  51. BaseTennot matchednot matchednot matchednot matchedno offer78 tok/s31.41.35s180 reqno offer
  52. BaseTennot matchednot matchednot matchednot matchedno offer94 tok/s35.91.25s134 reqno offer
  53. BaseTennot matchednot matchednot matchednot matchedno offer178 tok/s97632ms484 reqno offer
  54. DeepInfranot matchednot matchednot matchednot matchedno offer56 tok/s8859ms3 reqno offer
  55. DeepInfranot matchednot matchednot matchednot matchedno offer90 tok/s32792ms1,520 reqno offer
  56. Fireworksnot matchednot matchednot matchednot matchedno offer116 tok/s72542ms1,274 reqno offer
  57. Fireworksnot matchednot matchednot matchednot matchedno offer77 tok/s50890ms739 reqno offer
  58. Fireworksnot matchednot matchednot matchednot matchedno offer71 tok/s57427ms2,767 reqno offer
  59. Mistralnot matchednot matchednot matchednot matchedno offer88 tok/s48.9677ms200 reqno offer
  60. Mistralnot matchednot matchednot matchednot matchedno offer116 tok/s49752ms226 reqno offer
  61. Parasailnot matchednot matchednot matchednot matchedno offer136 tok/s60438ms2,728 reqno offer
  62. Parasailnot matchednot matchednot matchednot matchedno offer124 tok/s71.5384ms526 reqno offer
  63. Sail Researchnot matchednot matchednot matchednot matchedno offer61 tok/s37779ms560 reqno offer
  64. Sail Researchnot matchednot matchednot matchednot matchedno offer55 tok/s38661ms96 reqno offer
  65. Wafernot matchednot matchednot matchednot matchedno offer97 tok/s65289ms91,827 reqno offer
  66. Wafernot matchednot matchednot matchednot matchedno offer101 tok/s72207ms24,230 reqno offer

Each track runs from a route's p90, the rate nine in ten of its requests met or beat, up to its median. Weighted by requests, the figure above reports where the traffic went rather than what one route does: 25 of the 46 timed routes are slower than it, and the middle route runs at 91.5 tok/s. The widest median-to-p90 gap is DeepInfra's, 7x.

Alibaba publishes 2 offers and had 2 routes timed; BaseTen publishes 4 offers and had 4 routes timed; DeepInfra publishes 2 offers and had 2 routes timed; Fireworks publishes 3 offers and had 3 routes timed; Mistral publishes 2 offers and had 2 routes timed; Parasail publishes 2 offers and had 2 routes timed; Sail Research publishes 2 offers and had 2 routes timed; Wafer publishes 2 offers and had 2 routes timed. The two documents share no route id, so those offers and routes are printed apart rather than paired.

Uptime is OpenRouter's own figure for each seller over the trailing day, read once a day, and the column prints the latest reading, taken . 35 of the 61 sellers with a reading fell under 99% on at least one of the 23 days held, 23 Oct 2026 to 21 Nov 2026: Fireworks (fireworks/spot) at 0.00% on 2 Nov 2026, AtlasCloud at 71.79% on 4 Nov 2026, Crusoe at 78.43% on 15 Nov 2026, BaseTen (baseten/fp4) at 81.08% on 7 Nov 2026, and 31 more. Source: OpenRouter's public endpoints API, one reading a day for each seller: 1,211 sellers carried one on the latest of the 23 days held.

Source: OpenRouter's public endpoints API. 1257 sellers across 359 models when the mirror ran, 186 of them served by more than one. List prices, not what anybody paid; a discount is the seller's own. Read . Speed is OpenRouter's own timing over one 30-minute window, read .

What each seller accepts

21 parameters · 47 sellers
  1. include_reasoningevery seller
  2. max_tokensevery seller
  3. reasoningevery seller
  4. reasoning_effortevery seller
  5. temperatureevery seller
  6. toolsevery seller
  7. top_pevery seller
  8. tool_choice46 of 47not Amazon Bedrock
  9. response_format44 of 47not BaseTen (baseten/fast), BaseTen (baseten/us/fast), and Tenstorrent
  10. stop44 of 47not DigitalOcean, SiliconFlow, and Z.AI
  11. frequency_penalty41 of 47
  12. presence_penalty40 of 47
  13. structured_outputs38 of 47
  14. top_k36 of 47
  15. seed31 of 47
  16. repetition_penalty28 of 47
  17. logprobs24 of 47
  18. top_logprobs24 of 47
  19. logit_bias18 of 47
  20. min_p14 of 47
  21. parallel_tool_calls2 of 47only Cohere and Inceptron

7 of these 21 are accepted by every one of the 47 sellers that publish a list; the other 14 are declared by some sellers and not others, so what a request sending one of them gets depends on which seller serves it. Each list is the seller's own declaration in OpenRouter's endpoints API, read , and nothing here tested one.

Price position

342 priced models

Blended $0.523 per million tokens, 3 in to 1 out: cheaper than 210 of the 342 priced models in the catalog. Each bar is one stretch of the catalog on a logarithmic price scale; the mark is this listing.

Measured quality

145 models with an index · 132 on a design arena board

Each mark is one board this model stands on, placed at the share of that board it scores above, and led to its number in the list under the rule. The columns and the ledger below carry the figures.

Intelligence Index31.7Position 21 of 58, above 64% of the models scored.the intelligence leaderboard
Coding Index74.8Position 13 of 118, above 89% of the models scored.the coding leaderboard
Agentic Index53.1Position 5 of 94, above 95% of the models scored.the agentic leaderboard
  1. HTML slides · Agents1189 Elowin rate 39.1% · source rank 2the board, 25 ranked
  2. Mobile apps · Agents1184 Elowin rate 52.2% · source rank 25the board, 49 ranked
  3. python-pptx slides · Agents1217 Elowin rate 47.6% · source rank 10the board, 29 ranked
  4. Web apps · Agents1270 Elowin rate 54.2% · source rank 11the board, 47 ranked
  5. 3D scenes · Models1293 Elowin rate 55.5% · source rank 20the board, 113 ranked
  6. ASCII art · Models1184 Elowin rate 45.1% · source rank 29the board, 67 ranked
  7. Code · Models1282 Elowin rate 52.2% · source rank 22the board, 120 ranked
  8. Data visualisation · Models1251 Elowin rate 50.8% · source rank 34the board, 110 ranked
  9. Game development · Models1287 Elowin rate 50.4% · source rank 22the board, 120 ranked
  10. SVG · Models1244 Elowin rate 51.1% · source rank 20the board, 83 ranked
  11. UI components · Models1283 Elowin rate 53.1% · source rank 24the board, 116 ranked
  12. Websites · Models1278 Elowin rate 52.5% · source rank 24the board, 126 ranked
Epoch Capabilities Index · a fit by a third party over every benchmark this model has results onthe whole index, with intervals
Index157.1155.0 to 159.4
Position34 of 258above 87% of the models placed
Results behind it16benchmark results
As Epoch names itGLM-5.3released 14 Aug 2026

Position 34 of the 258 models Epoch has placed, which is not the population of either board above, on a scale anchored at Claude 3.5 Sonnet 130 and GPT-5 150. Epoch states that absolute values are meaningless by themselves and that comparisons between models are what the index is for, that values are not linearly related to benchmark accuracy, and that a highly specialised model may score low while being capable in its own domain.

Benchmark results

2 of 6 boards · 30 results
Epoch AI Benchmarking Hub · publisher-run, CC-BY-4.07 of 22 tasks · at max effort

Mean accuracy, 0 to 1, on the task the row names, as Epoch AI runs it and averaged over repeated runs. All eighteen of their tasks, including the held-out FrontierMath tiers, named as they name them. The model id carries the effort setting the run used. Epoch AI Benchmarking Hub names this run glm-5.3_max. All runs is how many runs the publisher scored on the task, every model's. Read from epoch.ai ↗, which dates its own data . Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.

LiveBench · publisher-run, Apache-2.023 of 23 tasks · 71 runs scored on each task · effort unstated

Percent correct on the task the row names, higher is better. All 23 published columns; there is no average to carry, because the site computes its headline in the browser from a weighting this does not reimplement. LiveBench names this run glm-5.3, with no effort setting written on it. Read from livebench.ai ↗, which dates its own data . LiveBench, Apache-2.0. Results produced by the LiveBench maintainers on questions refreshed monthly.

2 of the 6 openly licensed boards Vitna mirrors carry a result for GLM 5.3 (batch), each printed as its publisher published it and none added to another. MTEB, HELM, Vectara hallucination leaderboard and Aider polyglot have not published one under a name that matches its listing, and names are matched to the catalog exactly and never guessed. A task that is a link opens its column on /capability, with every run scored on it; an effort that is a link opens its run's row there, where the run is ranked on Vitna's scale. Every board, with its licence and what its publisher measures, is on /benchmarks.

Sources

Source: OpenRouter (openrouter.ai/models), as of 17 Nov 16:29 UTC. sha256 0bb53a77713f over 853,270 bytes.
Catalog: OpenRouter's public model list.
Intelligence, coding and agentic indexes: Artificial Analysis, as published by OpenRouter.
Elo, win rate and rank: Design Arena, as published by OpenRouter.
Source: each model's own Hugging Face repository (huggingface.co), as of 20 Nov 2026. 162 repositories read, 4 did not answer.
Licence, parameter count, model class and gating: each model's own Hugging Face repository, the one OpenRouter names for the listing.
A licence is the tag the uploader put on the repository, read from nothing else and verified by nobody.
A parameter count is the total the repository's safetensors files declare; for a mixture of experts that is every expert, not the share one token runs through.
Source: the Epoch AI Capabilities Index (epoch.ai/eci), as of 20 Nov 2026. 258 models placed over 2917 benchmark results.
Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.
Vitna ran none of these evaluations and adjusted none of these figures.