DeepSeekDeepseek v4 Flash Vision
Deepseek v4 Flash Vision is a DeepSeek proprietary reasoning model, released 21 Aug 2026. Its best configuration (max reasoning effort) ranks #158 of 785 on the BenchLeader Index at 58.2 ±4.9, in the upper half. The ± is the point: 258 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 agents & tools (70) and lowest in knowledge (54). At $0.660 per million tokens blended it is mid-priced. Output speed of 191 tokens per second puts it in the fastest quarter, with a first answer in 12.0 s. It has been measured at 2 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 21 Aug 2026.
- Blended price
- $0.660/M
- $0.440 in · $1.32 out
- Output speed
- 191 tok/s
- measured by Artificial Analysis
- First answer
- 12 s
- first token 1.07 s
- Context
- 1.0M
- Full answer
- 15 s
- median, reasoning included
- Cost per run
- $1.04
- one full Intelligence Index run
- Released
- 21 Aug 2026
- Overall index58
- Reasoning59
- Agents & tools70
- Knowledge54
- Multimodal58
- Long context66
- Composite69
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 |
|---|---|---|---|---|---|---|
| maxbest | 58.2 | #158 | 191 tok/s | 12 s | $0.0006 | Agents & tools 70 · Composite 69 · Knowledge 54 · Long context 66 · Multimodal 58 · Reasoning 59 |
| not stated | – | – | 90 tok/s | 1.07 s | $0.0002 | Composite 46 |
How its index has moved
19 Sept 2026 to 23 Oct 2026The 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
See all LLM benchmarks. 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 | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench 2.0 Reasoningnot in indexeffort not stated | – | 85.4%#31 | LiveBench | |
| LiveBench 1.0 Reasoningnot in indexeffort not stated | – | 85.4%#37 | LiveBench | |
| GPQA Diamond (AA)not in index | 91.3%#48 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 34.5%#127 | – | Artificial Analysis | |
| CritPt | 10.9%#115 | – | Artificial Analysis |
Coding
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench 2.0 Codingnot in indexeffort not stated | – | 37.6%#51 | LiveBench | |
| LiveBench 1.0 Codingnot in indexeffort not stated | – | 68.2%#64 | LiveBench | |
| SciCode (AA)not in index | 49.6%#114 | – | Artificial Analysis | |
| Terminal-Bench 4.0 (AA)not in index | 12.1%#84 | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 74.2%#66 | – | Artificial Analysis |
Agents & tools
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench Long Agenticnot in indexeffort not stated | – | 39.0%#26 | LiveBench | |
| LiveBench Browsernot in indexeffort not stated | – | 64.0%#26 | LiveBench | |
| LiveBench Agentic Codingnot in indexeffort not stated | – | 65.1%#6 | LiveBench | |
| GDPval-AA v2.1not in index | 52.3%#32 | – | Artificial Analysis | |
| τ³-Banking (AA) | 41.0%#27 | – | Artificial Analysis | |
| AutomationBenchnot in index | 47.5%#87 | – | Artificial Analysis | |
| GDP.pdf (AA)not in index | 12.0%#116 | – | Artificial Analysis | |
| EnterpriseOps-Gymnot in index | 47.1%#11 | – | Artificial Analysis |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| LiveBench Mathematicsnot in index | 87.8%#45 | LiveBench |
Knowledge
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench 2.0 Data Analysisnot in indexeffort not stated | – | 79.5%#12 | LiveBench | |
| LiveBench Researchnot in indexeffort not stated | – | 69.2%#35 | LiveBench | |
| LiveBench 1.0 Data Analysisnot in indexeffort not stated | – | 79.5%#13 | LiveBench | |
| AA-Omniscience | -17.6#233 | – | Artificial Analysis | |
| AA-Omniscience: accuracynot in index | 38.6%#133 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 8.5%#457 | – | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| LiveBench Languagenot in index | 80.4%#29 | LiveBench | |
| LiveBench Instruction Followingnot in index | 71.0%#26 | LiveBench |
Multimodal
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench Designnot in indexeffort not stated | – | 70.7%#38 | LiveBench | |
| MMMU-Pro | 74.8%#108 | – | Artificial Analysis |
Long context
| Benchmark | max | Source | Trend |
|---|---|---|---|
| AA-LCR | 81.3%#56 | Artificial Analysis |
Composite
| Benchmark | max | not stated | Source | Trend |
|---|---|---|---|---|
| LiveBench 2.0effort not stated | – | 63.6%#34 | LiveBench | |
| LiveBench 1.0not in indexeffort not stated | – | 76.8%#28 | LiveBench | |
| AA Intelligence Index v4.3.2 | 34.8#90 | – | Artificial Analysis |
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 |
|---|---|---|---|---|---|---|
| GMICloud | 95 tok/s | 1.74 s | $0.440 | $1.32 | 1.0M | fp8 |
| DeepInfra | 68 tok/s | 1.01 s | $0.216 | $0.647 | 1.0M | fp8 |
| SiliconFlow | 58 tok/s | 1.30 s | $0.440 | $1.32 | 1.0M | fp8 |
Price history
The price above, per 1M tokens, every day it changed.
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 $0.003 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0006 | $0.0004 | 13.6 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0061 | $0.0021 | 15.1 s |
| Code edit | 6,000 / 1,500 | $0.0046 | $0.0027 | 19.8 s |
| Agentic coding session | 60,000 / 4,000 | $0.032 | $0.012 | 33.0 s |
| Structured extraction | 2,000 / 200 | $0.0011 | $0.0005 | 13.0 s |
More from DeepSeek
See also
Questions people ask
- What are Deepseek v4 Flash Vision's benchmark scores?
- As of 24 Oct 2026, Deepseek v4 Flash Vision (max effort) scores 58.2 ±4.9 on the BenchLeader Index, strongest in agents & tools (70) and weakest in knowledge (54). That rests on 6 independent benchmark results, and every score on this page links to its publisher.
- How does Deepseek v4 Flash Vision rank?
- Deepseek v4 Flash Vision (max effort) is #158 of 785 ranked configurations on 24 Oct 2026. 258 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Deepseek v4 Flash Vision cost per 1M tokens?
- As of 24 Oct 2026, Deepseek v4 Flash Vision costs $0.440 per 1M input tokens and $1.32 per 1M output tokens, $0.660 blended (three input tokens per output token). The list price last changed on 6 Oct 2026.
- When was Deepseek v4 Flash Vision released?
- DeepSeek released Deepseek v4 Flash Vision on 21 Aug 2026. BenchLeader has results for it at 2 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Deepseek v4 Flash Vision: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/deepseek-v4-flash-vision.