DeepSeekDeepSeek V4 Flash 0731
DeepSeek V4 Flash 0731 is a DeepSeek open-weights reasoning model, released 31 Jul 2026. Its best configuration (high reasoning effort) ranks #235 of 751 on the BenchLeader Index at 54.4 ±5.3, in the upper half. The ± is the point: 297 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in reasoning (59) and lowest in coding (54). At $0.175 per million tokens blended it is among the cheapest fifth of ranked models. Output speed of 35 tokens per second puts it in the slowest quarter, with a first token in 0.8 s. It has been measured at 5 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 29 Sept 2026.
- Blended price
- $0.175/M
- $0.140 in · $0.280 out
- Output speed
- 35 tok/s
- OpenRouter traffic, 7-day median; not yet measured by Artificial Analysis
- First answer
- 0.76 s
- Context
- 1.0M
- Full answer
- –
- Cost per run
- –
- Released
- 31 Jul 2026
- Overall index54
- Reasoning59
- Coding54
- Knowledge55
Versions
DeepSeek has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| DeepSeek V4 Flash 0731highthis page | 31 Jul 2026 | 54.4 | #235 |
| DeepSeek V4 Flashmax | 24 Apr 2026 | 60.7 | #100 |
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | – | – | 35 tok/s | 0.76 s | $0.0001 | Reasoning 31 |
| low | – | – | 35 tok/s | 0.76 s | $0.0001 | Reasoning 56 |
| highbest | 54.4 | #235 | 35 tok/s | 0.76 s | $0.0001 | Coding 54 · Knowledge 55 · Reasoning 59 |
| max | 53.5 | #269 | 35 tok/s | 0.76 s | $0.0001 | Coding 57 · Knowledge 43 · Maths 56 · Reasoning 63 |
| not stated | 50.1 | #357 | 35 tok/s | 0.76 s | $0.0001 | Agents & tools 55 · Coding 31 · Composite 47 · Maths 59 · Reasoning 60 |
How its index has moved
10 Sept 2026 to 2 Oct 2026- high
- max
- not stated
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | – | 91.0%#31 | – | Epoch AI Benchmarking Hub | |
| SimpleBench | – | – | – | – | 61.1%#26 | SimpleBench | |
| LiveBench Reasoningnot in index | – | – | – | – | 86.6%#31 | LiveBench | |
| GPQA Diamond (Vals)not in index | – | – | 89.9%#27 | – | – | Vals AI | |
| ARC-AGI-1 | 11.8%#198 | 84.0%#96 | 87.0%#81 | 89.0%#72 | – | ARC Prize | |
| ARC-AGI-2 | 2.1%#171 | 46.0%#96 | 56.0%#84 | 61.4%#71 | – | ARC Prize | |
| Chess Puzzlesnot in index | – | – | – | 33.0%#45 | – | Epoch AI Benchmarking Hub | |
| Mystery Game Puzzlesnot in index | – | – | – | 34.0%#21 | – | Epoch AI Benchmarking Hub | |
| LMCAnot in index | – | – | – | 41.7%#82 | – | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | – | – | – | 90.9%#60 | – | Epoch AI Benchmarking Hub |
Coding
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| SciCode | – | – | – | 49.9%#89 | – | SciCode | |
| WeirdML | – | – | 57.0%#55 | 63.0%#42 | – | WeirdML | |
| FrontierCode | – | – | – | – | 18.8%#34 | Cognition | |
| LiveBench Codingnot in index | – | – | – | – | 75.0%#48 | LiveBench | |
| LiveCodeBench | – | – | 87.3%#16 | – | – | Vals AI | |
| IOI | – | – | 32.7%#33 | – | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | 88.8%#11 | – | – | Vals AI | |
| Code Migrationnot in index | – | – | 38.6%#27 | – | – | Vals AI | |
| ProgramBenchnot in index | – | – | 0.0%#23 | – | – | Vals AI | |
| Vibe Code Bench v1.1not in index | – | – | 74.7%#28 | – | – | Vals AI | |
| ALE-Benchnot in index | – | – | – | 1306.1#27 | – | Epoch AI Benchmarking Hub |
Agents & tools
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| LiveBench Agentic Codingnot in index | – | – | – | – | 46.8%#46 | LiveBench | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | 67.0%#36 | Vals AI | |
| CyberBenchnot in index | – | – | 72.1%#14 | – | – | Vals AI | |
| Finance Agent v2not in index | – | – | 49.5%#39 | – | – | Vals AI | |
| Harvey's Legal Agent Benchmarknot in index | – | – | 8.3%#19 | – | – | Vals AI | |
| Legal Research Benchnot in index | – | – | 30.3%#38 | – | – | Vals AI | |
| SkillsBenchnot in index | – | – | 50.7%#21 | – | – | Vals AI | |
| Tax Agent Benchnot in index | – | – | 62.6%#24 | – | – | Vals AI | |
| Terminal-Bench 4.0 (Vals)not in index | – | – | 18.7%#23 | – | – | Vals AI | |
| Terminal-Bench Sciencenot in index | – | – | 0.0%#29 | – | – | Vals AI |
Maths
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | 57.5%#37 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | 24.4%#43 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | – | – | 94.4%#46 | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | 56.0%#18 | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | 86.8%#46 | LiveBench |
Knowledge
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | 33.6%#54 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | 79.3%#13 | LiveBench | |
| MMLU-Pro | – | – | 86.2%#45 | – | – | Vals AI | |
| LegalBench | – | – | 77.7%#102 | – | – | Vals AI | |
| CorpFin | – | – | 61.9%#52 | – | – | Vals AI | |
| TaxEval | – | – | 70.7%#84 | – | – | Vals AI | |
| BioMysteryBenchnot in index | – | – | 64.4%#15 | – | – | Vals AI | |
| Excel Modeling Benchmarknot in index | – | – | 57.0%#37 | – | – | Vals AI | |
| MedCodenot in index | – | – | 41.4%#50 | – | – | Vals AI | |
| MedScribenot in index | – | – | 80.4%#53 | – | – | Vals AI |
Instruction following
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 79.2%#36 | LiveBench | |
| LiveBench Instruction Followingnot in index | – | – | – | – | 65.5%#41 | LiveBench |
Composite
| Benchmark | no reasoning | low | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | 154.5#31 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | 74.2%#41 | LiveBench | |
| Vals Indexnot in index | – | – | 48.0#29 | – | – | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Baseten | 185 tok/s | 0.48 s | $0.130 | $0.260 | 1.0M | fp8 |
| Baseten | 174 tok/s | 0.41 s | $0.130 | $0.260 | 1.0M | fp8 |
| Cohere | 174 tok/s | 0.39 s | $0.140 | $0.280 | 1.0M | – |
| CoreWeave | 121 tok/s | 0.49 s | $0.130 | $0.280 | 262k | fp8 |
| Parasail | 99 tok/s | 0.77 s | $0.140 | $0.280 | 1.0M | fp8 |
| Baidu Qianfan | 96 tok/s | 0.68 s | $0.440 | $1.32 | 1.0M | fp8 |
| Wafer | 95 tok/s | 0.48 s | $0.120 | $0.840 | 1.0M | – |
| Reka AI | 94 tok/s | 0.58 s | $0.021 | $0.528 | 262k | – |
| AtlasCloud | 89 tok/s | 1.22 s | $0.440 | $1.32 | 1.0M | fp4 |
| Inceptron | 87 tok/s | 0.62 s | $0.050 | $0.650 | 1.0M | fp4 |
| Alibaba Cloud Int. | 80 tok/s | 1.38 s | $0.352 | $1.06 | 1M | – |
| NextBit | 73 tok/s | 2.51 s | $0.352 | $1.06 | 1.0M | fp8 |
| NovitaAI | 71 tok/s | 1.96 s | $0.409 | $1.23 | 1.0M | fp8 |
| Mancer | 70 tok/s | 0.97 s | $0.200 | $0.600 | 1.0M | fp8 |
| GMICloud | 65 tok/s | 1.70 s | $0.286 | $0.858 | 1.0M | fp8 |
| Venice | 60 tok/s | 0.93 s | $0.175 | $0.350 | 1M | – |
| SiliconFlow | 57 tok/s | 1.41 s | $0.220 | $0.660 | 1.0M | fp8 |
| Cloudflare | 55 tok/s | 1.12 s | $0.440 | $1.32 | 1.0M | – |
| DeepInfra | 46 tok/s | 0.76 s | $0.060 | $0.180 | 1.0M | fp8 |
| Together | 45 tok/s | 0.60 s | $0.140 | $0.280 | 1.0M | – |
| Phala | 44 tok/s | 0.81 s | $0.308 | $0.924 | 1.0M | – |
| StreamLake | 40 tok/s | 1.26 s | $0.044 | $0.132 | 1.0M | fp8 |
| DigitalOcean | 39 tok/s | 0.51 s | $0.119 | $0.238 | 1.0M | – |
| Morph | 36 tok/s | 0.78 s | $0.142 | $0.400 | 1.0M | bf16 |
| Open Inference | 35 tok/s | 1.12 s | $0.020 | $1.66 | 1.0M | fp4 |
| Sail Research | 32 tok/s | 1.80 s | $0.019 | $0.420 | 1.0M | fp4 |
| Sail Research | 30 tok/s | 1.93 s | $0.019 | $0.300 | 1.0M | fp4 |
| Relace | 27 tok/s | 1.08 s | $0.015 | $1.28 | 1.0M | fp4 |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.030 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0001 | $0.0001 | 9.3 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0018 | $0.0009 | 17.9 s |
| Code edit | 6,000 / 1,500 | $0.0013 | $0.0008 | 43.6 s |
| Agentic coding session | 60,000 / 4,000 | $0.0095 | $0.0046 | 1.9 min |
| Structured extraction | 2,000 / 200 | $0.0003 | $0.0002 | 6.5 s |
See also
Data as of 4 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “DeepSeek V4 Flash 0731: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/deepseek-v4-flash-0731, data as of 4 Oct 2026.