DeepSeek V4.1 Flash 1017
Sold by 51 sellers, timed at 148 tok/s, scored 28.6 on the Intelligence Index, on 9 Design Arena boards, and with results on 1 openly licensed board.
retrieved sha256 · 0bb53a77713f
The same weights served by 4 providers, each at its own list price. OpenRouter's rate is its routed rate across whichever of them it uses, and it need not match any one of them.
What sellers charge
51 sellers · 20x from cheapest to dearestOne tick per seller, on a log scale of input price per million from $0.03 at Particle AI to $0.60 at Morph. 1 seller undercuts the $0.036 listed; 49 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 not declared 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
2,266,465 requests · 51 endpoints · 30-minute windowOpenRouter'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
51 priced · 51 timed · cheapest first- Particle AIparticle-ai/us/flex$0.03$0.275$0.015not declared131knot matchednonenonenot matched95.76%
- Waferwafer/us · listed$0.036$0.50$0.01mxfp4944knot matchednonenonenot matched100%
- Sail Researchsail-research/fp4$0.04$0.10$0.007fp4384k121 tok/s57938ms69,063 req99.77%
- Sail Researchsail-research/fp4/us$0.04$0.10$0.007fp4384k122 tok/s511.19s30,381 req99.77%
- Ambientambient/fp4$0.045$0.25$0.0025fp4944k35 tok/s123.09s3,860 req99.51%
- OpenInferenceopen-inference/fp4$0.05$0.75$0.002fp4944k170 tok/s77860ms55,089 req99.86%
- Pearl Researchpearl-research/fp8$0.05$0.30$0.003fp8944k138 tok/s541.18s103,074 req100%
- PrimeIntellectprimeintellect$0.056784$0.70$0.00312not declared131k124 tok/s69612ms102,903 req99.86%
- Morphmorph/fp8$0.059$0.52$0.005fp8944knot matchednonenonenot matched98.33%
- Particle AIparticle-ai/us$0.06$0.55$0.03not declared131knot matchednonenonenot matched100%
- Waferwafer/mxfp4$0.066$0.50$0.0065mxfp4944knot matchednonenonenot matched100%
- StreamLakestreamlake/fp8 · 75% off$0.075$0.30$0.0015fp8384k68 tok/s331.26s2,545 req100%
- DeepInfradeepinfra/fp8/flex · 51% off$0.0784$0.2352$0.002352fp8393knot matchednonenonenot matched100%
- InferenceNetinference-net/us$0.08$0.666$0.008fp8393k157 tok/s89349ms37,737 req100%
- InferenceNetinference-net/fp8$0.08$0.666$0.008fp8393k160 tok/s95207ms125,571 req100%
- Alibabaalibaba$0.09$0.36$0.0018not declared393k141 tok/s65886ms47,343 req100%
- Baidubaidu/fp8 · 65% off$0.09772$0.39437$0.002094fp8393k130 tok/s461.08s14,535 req100%
- DeepInfradeepinfra/fp8 · 51% off$0.098$0.294$0.00294fp8393knot matchednonenonenot matched100%
- Cloudflarecloudflare$0.10$0.60$0.003not declared944k96 tok/s362.06s11,787 req100%
- DekaLLMdekallm$0.10$0.60$0.003not declared393k66 tok/s261.52s3,055 req100%
- Inceptroninceptron/fp4$0.10$0.60$0.006fp4944k146 tok/s79.2366ms193 req99.00%
- StepFunstepfun/fp8 · 65% off$0.105$0.42$0.0021fp8384k161 tok/s65995ms34,143 req100%
- Coherecohere/fp8$0.115$0.40$0.0025fp8384k148 tok/s66859ms67,593 req100%
- Ant Digitalantdigital/fp8 · 60% off$0.12$0.48$0.0024fp8393k109 tok/s45.21.75s1,101 req100%
- Relacerelace/fp4$0.12$0.50$0.003fp4944k150 tok/s65974ms59,151 req99.93%
- Riverriver$0.12$0.48$0.0024not declared393k82 tok/s51546ms17,890 req100%
- DeepInfradeepinfra/fp8/priority · 51% off$0.147$0.441$0.00441fp8393knot matchednonenonenot matched100%
- DeepSeekdeepseek$0.15$0.60$0.003not declared393k128 tok/s581.06s297,822 req100%
- DigitalOceandigitalocean$0.15$0.60$0.003not declared944k187 tok/s84752ms59,324 req100%
- Fireworksfireworks/spot$0.15$0.60$0.003not declared944knot matchednonenonenot matched96.29%
- Makoramakora/fp8$0.15$0.80$0.006fp8944k129 tok/s58874ms7,429 req100%
- SiliconFlowsiliconflow/fp8$0.15$0.60$0.003fp8393k153 tok/s59826ms24,183 req100%
- Novitanovita/fp8 · 45% off$0.165$0.66$0.0033fp8393k82 tok/s331.42s3,827 req100%
- ModelRunmodelrun/fp4$0.17$0.67$0.006fp4131k158 tok/s69794ms12,575 req100%
- Nebiusnebius/fp4 · 40% off$0.18$0.72$0.0036fp4944k238 tok/s78694ms78,183 req100%
- Ionstreamionstream$0.21$0.62$0.0105not declared944k214 tok/s95552ms23,275 req99.86%
- PrimaLabsprimalabs$0.24$0.96$0.0048not declared944k134 tok/s51607ms23 req97.88%
- Modalmodal$0.25$1.00$0.005not declared944k169 tok/s89625ms2,821 req100%
- Crusoecrusoe/fp8$0.29$1.20$0.007fp8944k188 tok/s85593ms11,618 req100%
- BaseTenbaseten$0.30$1.20$0.007not declared262knot matchednonenonenot matched100%
- CoreWeavecoreweave/fp8$0.30$1.20$0.006fp8944k228 tok/s127437ms65,322 req100%
- Fireworksfireworks$0.30$1.20$0.006not declared944knot matchednonenonenot matched100%
- Mistralmistral/fp8$0.30$1.20$0.007fp8944k148 tok/s481.29s3,684 req99.13%
- Near AInear-ai$0.30$1.20$0.006not declared944k46 tok/s19.1694ms26 req100%
- Parasailparasail/fp8$0.30$1.20$0.006fp8944k199 tok/s101572ms12,696 req100%
- Rekareka$0.30$1.20$0.006not declared393k154 tok/s81833ms3,580 req99.93%
- Togethertogether$0.30$1.20$0.006not declared944k186 tok/s84615ms36,826 req100%
- Venicevenice/fp8$0.30$1.20$0.0075fp8262k119 tok/s46.41.56s196 req99.66%
- Fireworksfireworks/us$0.45$1.80$0.009not declared944knot matchednonenonenot matched100%
- BaseTenbaseten/fast$0.60$2.40$0.014not declared262knot matchednonenonenot matched100%
- Morphmorph/fast$0.60$2.40$0.014fp8944knot matchednonenonenot matched100%
- BaseTennot matchednot matchednot matchednot matchedno offer538 tok/s199341ms8,601 reqno offer
- BaseTennot matchednot matchednot matchednot matchedno offer160 tok/s101478ms16,154 reqno offer
- DeepInfranot matchednot matchednot matchednot matchedno offer125 tok/s68981ms39,117 reqno offer
- DeepInfranot matchednot matchednot matchednot matchedno offer115 tok/s66758ms289,964 reqno offer
- DeepInfranot matchednot matchednot matchednot matchedno offer96 tok/s312.42s27,388 reqno offer
- Fireworksnot matchednot matchednot matchednot matchedno offer118 tok/s531.2s24,449 reqno offer
- Fireworksnot matchednot matchednot matchednot matchedno offer158 tok/s66812ms1,404 reqno offer
- Fireworksnot matchednot matchednot matchednot matchedno offer185 tok/s81.5600ms4,188 reqno offer
- Morphnot matchednot matchednot matchednot matchedno offer241 tok/s115488ms6,519 reqno offer
- Morphnot matchednot matchednot matchednot matchedno offer121 tok/s76460ms67,102 reqno offer
- Particle AInot matchednot matchednot matchednot matchedno offer114 tok/s43778ms121,389 reqno offer
- Particle AInot matchednot matchednot matchednot matchedno offer105 tok/s56496ms6,501 reqno offer
- Wafernot matchednot matchednot matchednot matchedno offer184 tok/s99319ms57,046 reqno offer
- Wafernot matchednot matchednot matchednot matchedno offer179 tok/s96349ms166,219 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: 26 of the 51 timed routes are slower than it, and the middle route runs at 146 tok/s. The widest median-to-p90 gap is DeepInfra's, 3.1x.
BaseTen publishes 2 offers and had 2 routes timed; DeepInfra publishes 3 offers and had 3 routes timed; Fireworks publishes 3 offers and had 3 routes timed; Morph publishes 2 offers and had 2 routes timed; Particle AI 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 . 36 of the 63 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 4 Nov 2026, Sail Research (sail-research/fp4) at 0.00% on 8 Nov 2026, Sail Research (sail-research/fp4/us) at 0.00% on 8 Nov 2026, DeepInfra (deepinfra/fp8/flex) at 58.33% on 14 Nov 2026, and 32 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
22 parameters · 51 sellers- include_reasoningevery seller
- max_tokensevery seller
- reasoningevery seller
- reasoning_effortevery seller
- tool_choiceevery seller
- toolsevery seller
- temperature50 of 51not Near AI
- top_p50 of 51not Near AI
- response_format49 of 51not BaseTen (baseten) and BaseTen (baseten/fast)
- stop48 of 51not DigitalOcean, Near AI, and SiliconFlow
- structured_outputs46 of 51
- frequency_penalty44 of 51
- presence_penalty41 of 51
- logprobs35 of 51
- top_logprobs35 of 51
- top_k34 of 51
- seed32 of 51
- repetition_penalty28 of 51
- logit_bias24 of 51
- min_p18 of 51
- parallel_tool_calls3 of 51only Cohere, Inceptron, and PrimeIntellect
- max_completion_tokens1 of 51only PrimeIntellect
6 of these 22 are accepted by every one of the 51 sellers that publish a list; the other 16 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 modelsBlended $0.152 per million tokens, 3 in to 1 out: cheaper than 282 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 boardEach 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.
- spreadsheet · Agents1265 Elowin rate 47.8% · source rank 9the board, 32 ranked
- 3D scenes · Models1344 Elowin rate 47.9% · source rank 13the board, 113 ranked
- ASCII art · Models1219 Elowin rate 47.8% · source rank 23the board, 67 ranked
- Code · Models1286 Elowin rate 48.4% · source rank 20the board, 120 ranked
- Data visualisation · Models1276 Elowin rate 51.5% · source rank 22the board, 110 ranked
- Game development · Models1253 Elowin rate 43.2% · source rank 35the board, 120 ranked
- SVG · Models1241 Elowin rate 50.7% · source rank 21the board, 83 ranked
- UI components · Models1280 Elowin rate 48.6% · source rank 27the board, 116 ranked
- Websites · Models1293 Elowin rate 51.6% · source rank 18the board, 126 ranked
Position 24 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
1 of 6 boards · 7 results| Task | All runs | Score |
|---|---|---|
| Chess Puzzles | 167 | 0.46 |
| FrontierMath-Tier-4-v2-Private | 84 | 0.46 |
| FrontierMath-Tiers-1-3-v2-Private | 128 | 0.71 |
| GPQA diamond | 342 | 0.93 |
| Mystery Game Puzzles | 149 | 0.49 |
| OTIS Mock AIME 2024-2025 | 319 | 0.99 |
| SimpleQA Verified | 106 | 0.42 |
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 deepseek-v4.1-flash-1017_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.
1 of the 6 openly licensed boards Vitna mirrors carries a result for DeepSeek V4.1 Flash 1017, each printed as its publisher published it and none added to another. MTEB, LiveBench, 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.
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.
