Get started

DeepSeek V4.1 Flash

Sold by 37 sellers, timed at 150 tok/s, with results on 2 openly licensed boards, and not scored by Artificial Analysis.

retrieved sha256 · 0bb53a77713f

Models/DeepSeek/DeepSeek V4.1 FlashUSD per 1M tokens · as OpenRouter routes it
deepseek/deepseek-v4.1-flashDeepSeek · base listing · DeepSeek tokenizer
OpenRouter list price · the routed rate, as the catalog carried itretrieved 17 Nov 16:29 UTC from openrouter.ai ↗
Input$0.30per 1M tokens
Output$1.20per 1M tokens
Cached input$0.006per 1M tokens
Context1.05M1,048,576 tokens
What Vitna read · 5 providers, each from its own pricing page, kept as dated recordsin / out per 1M tokens
Novita AIopen weights$0.30 / $1.20confirmed 22 Nov 2026
Basetenopen weights$0.30 / $1.20confirmed 22 Nov 2026
Parasailopen weights$0.30 / $1.20confirmed 22 Nov 2026
Together AIopen weights$0.30 / $1.20confirmed 22 Nov 2026
Basetenopen weights$0.60 / $2.40confirmed 22 Nov 2026

The same weights served by 5 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 the list says · the fields OpenRouter publishes for this id11 fields
AuthorDeepSeek
Listed
Canonical iddeepseek/deepseek-v4.1-flash-20260910
Context1,048,576 tokens
Max output943,718 tokens
Modalitiestext, image → text
TokenizerDeepSeek
Reasoningoptional, on by default · effort max, high, low, none, high by default
Moderationnot moderated by OpenRouter
Other listingbatch variant
What the weights say · read from the model's own repository on Hugging Faceretrieved from huggingface.co ↗
Licencemit
Parameters763B · 763,205,315,794 declared
Model classDeepseekV41ForCausalLM
Downloadnot gated
Published

What sellers charge

37 sellers · 30.12x from cheapest to dearest

One tick per seller, on a log scale of input price per million from $0.0498 at Baidu to $1.50 at Venice. 24 sellers undercut the $0.30 listed; 3 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 fp8 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

1,389,152 requests · 37 endpoints · 30-minute window
Throughput150 tok/sMedian output rate, weighted across the 37 endpoints serving it by the requests each answered.the speed board
First token618msMedian wait before the first token, over the same window and weighted the same way.how Vitna reads a figure
Endpoint spread84 to 271The 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

37 priced · 37 timed · cheapest first
  1. Baidubaidu/fp8 · 83% off$0.0498$0.1992$0.000996fp8393k109 tok/s411.35s276,793 req100%
  2. OpenInferenceopen-inference/fp4$0.05$0.75$0.002fp4944k168 tok/s76607ms38,131 req99.93%
  3. Waferwafer/us$0.051$0.567$0.018mxfp4944knot matchednonenonenot matched100%
  4. Alibabaalibaba$0.09$0.36$0.0018not declared393k145 tok/s70716ms46,012 req100%
  5. Cloudflarecloudflare$0.10$0.90$0.03not declared944k88 tok/s351.33s647 req98.84%
  6. InferenceNetinference-net/fp8$0.103$0.60$0.008fp8393k180 tok/s103217ms74,603 req100%
  7. InferenceNetinference-net/us$0.103$0.60$0.008fp8393k153 tok/s92228ms65,024 req100%
  8. DeepInfradeepinfra/fp8/flex · 30% off$0.112$0.336$0.00336fp8393knot matchednonenonenot matched95.41%
  9. Rekareka · 20% off$0.12$0.64$0.008not declared393k107 tok/s68201ms27,179 req100%
  10. Relacerelace/fp4$0.12$0.50$0.003fp4944k140 tok/s67777ms24,526 req99.93%
  11. Riverriver$0.12$0.48$0.0024not declared393k146 tok/s76509ms41,954 req100%
  12. CoreWeavecoreweave/fp8$0.13$0.65$0.002fp8944k191 tok/s89560ms64,661 req100%
  13. DeepInfradeepinfra/fp8 · 30% off$0.14$0.42$0.0042fp8393knot matchednonenonenot matched100%
  14. StreamLakestreamlake/fp8 · 51% off$0.147$0.588$0.00294fp8384k84 tok/s45.1959ms1,181 req100%
  15. Fireworksfireworks/spot$0.15$0.60$0.003not declared944knot matchednonenonenot matched99.58%
  16. Novitanovita/fp8 · 50% off$0.15$0.60$0.003fp8393k110 tok/s65975ms4,344 req100%
  17. Waferwafer/mxfp4$0.19$0.65$0.012mxfp4944knot matchednonenonenot matched100%
  18. Coherecohere/fp8$0.195$0.66$0.003fp8384k188 tok/s92580ms42,058 req100%
  19. DigitalOceandigitalocean$0.195$0.78$0.0039not declared944k177 tok/s84642ms29,182 req100%
  20. Ionstreamionstream$0.20$0.70$0.02not declared944k145 tok/s78191ms18,079 req100%
  21. DeepInfradeepinfra/fp8/priority · 30% off$0.21$0.63$0.0063fp8393knot matchednonenonenot matched100%
  22. Mancer 2mancer/fp8$0.2125$0.75$0.07fp8944k127 tok/s70789ms593 req100%
  23. Makoramakora/fp8$0.26$1.17$0.0058fp8944k154 tok/s82778ms3,866 req100%
  24. Crusoecrusoe/fp8$0.29$1.20$0.007fp8944k207 tok/s106467ms15,655 req100%
  25. BaseTenbaseten/us · listed$0.30$1.20$0.007not declared262k170 tok/s78212ms1,197 req100%
  26. BaseTenbaseten · listed$0.30$1.20$0.007not declared262k257 tok/s148548ms2,326 req100%
  27. Fireworksfireworks · listed$0.30$1.20$0.006not declared944knot matchednonenonenot matched100%
  28. Googlegoogle-vertex/global · listed$0.30$1.20$0.03not declared944k98 tok/s51589ms5,413 req100%
  29. Modalmodal · listed$0.30$1.20$0.03not declared944k180 tok/s103402ms337 req100%
  30. Nebiusnebius/fp8 · listed$0.30$1.20$0.01fp8944k180 tok/s63806ms476 req100%
  31. Parasailparasail/fp8 · listed$0.30$1.20$0.006fp8944knot matchednonenonenot matched100%
  32. SiliconFlowsiliconflow/fp8 · listed$0.30$1.20$0.006fp8393k122 tok/s611.58s1,215 req99.79%
  33. Togethertogether · listed$0.30$1.20$0.006not declared944k271 tok/s128216ms88,588 req100%
  34. Venicevenice/fp8 · listed$0.30$1.20$0.0075fp8262knot matchednonenonenot matched99.93%
  35. Fireworksfireworks/us$0.45$1.80$0.009not declared944knot matchednonenonenot matched100%
  36. Parasailparasail/fast$0.45$1.80$0.009fp8944knot matchednonenonenot matched100%
  37. Venicevenice/ultrafast$1.50$6.00$0.03fp8131knot matchednonenonenot matched100%
  38. DeepInfranot matchednot matchednot matchednot matchedno offer113 tok/s74375ms34,278 reqno offer
  39. DeepInfranot matchednot matchednot matchednot matchedno offer114 tok/s412.02s18,219 reqno offer
  40. DeepInfranot matchednot matchednot matchednot matchedno offer110 tok/s67491ms241,665 reqno offer
  41. Fireworksnot matchednot matchednot matchednot matchedno offer160 tok/s106270ms565 reqno offer
  42. Fireworksnot matchednot matchednot matchednot matchedno offer224 tok/s129315ms2,812 reqno offer
  43. Fireworksnot matchednot matchednot matchednot matchedno offer113 tok/s67415ms49,482 reqno offer
  44. Parasailnot matchednot matchednot matchednot matchedno offer207 tok/s104399ms11,547 reqno offer
  45. Parasailnot matchednot matchednot matchednot matchedno offer179 tok/s117328ms2,792 reqno offer
  46. Venicenot matchednot matchednot matchednot matchedno offer233 tok/s85.1503ms5 reqno offer
  47. Venicenot matchednot matchednot matchednot matchedno offer187 tok/s89.8398ms2,895 reqno offer
  48. Wafernot matchednot matchednot matchednot matchedno offer196 tok/s107113ms78,599 reqno offer
  49. Wafernot matchednot matchednot matchednot matchedno offer184 tok/s101271ms72,253 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: 16 of the 37 timed routes are slower than it, and the middle route runs at 160 tok/s. The widest median-to-p90 gap is Nebius's, 2.85x.

DeepInfra publishes 3 offers and had 3 routes timed; Fireworks publishes 3 offers and had 3 routes timed; Parasail publishes 2 offers and had 2 routes timed; Venice 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 . 29 of the 50 sellers with a reading fell under 99% on at least one of the 23 days held, 23 Oct 2026 to 21 Nov 2026: DeepInfra (deepinfra/fp8/priority) at 0.63% on 24 Oct 2026, DeepInfra (deepinfra/fp8/flex) at 28.89% on 10 Nov 2026, Mancer 2 at 63.71% on 24 Oct 2026, Decart at 83.33% on 26 Oct 2026, and 25 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 · 37 sellers
  1. include_reasoningevery seller
  2. max_tokensevery seller
  3. reasoningevery seller
  4. reasoning_effortevery seller
  5. temperatureevery seller
  6. tool_choiceevery seller
  7. toolsevery seller
  8. top_pevery seller
  9. response_format36 of 37not BaseTen (baseten)
  10. stop35 of 37not DigitalOcean and SiliconFlow
  11. structured_outputs34 of 37not Alibaba, SiliconFlow, and StreamLake
  12. frequency_penalty32 of 37
  13. presence_penalty31 of 37
  14. top_k30 of 37
  15. repetition_penalty23 of 37
  16. seed23 of 37
  17. logit_bias21 of 37
  18. logprobs19 of 37
  19. top_logprobs19 of 37
  20. min_p14 of 37
  21. parallel_tool_calls3 of 37only BaseTen (baseten), BaseTen (baseten/us), and Cohere
  22. top_a1 of 37only Mancer 2

8 of these 22 are accepted by every one of the 37 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 $0.525 per million tokens, 3 in to 1 out: cheaper than 206 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
Intelligence Indexnot scoredArtificial Analysis has not scored this model on its Intelligence Index.the intelligence leaderboard
Coding Indexnot scoredArtificial Analysis has not scored this model on its Coding Index.the coding leaderboard
Agentic Indexnot scoredArtificial Analysis has not scored this model on its Agentic Index.the agentic leaderboard

Not ranked in any Design Arena category.

Epoch Capabilities Index · a fit by a third party over every benchmark this model has results onthe whole index, with intervals
Index157.0155.1 to 159.2
Position35 of 258above 87% of the models placed
Results behind it20benchmark results
As Epoch names itDeepSeek V4.1 Flashreleased 9 Sep 2026

Position 35 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 · 32 results
Epoch AI Benchmarking Hub · publisher-run, CC-BY-4.09 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 deepseek-v4.1-flash_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 · at max effort

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 deepseek-v4.1-flash-max. 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 DeepSeek V4.1 Flash, 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.