DeepSeek-V4-Flash-0731: benchmarks, price and rank
DeepSeek-V4-Flash-0731 from DeepSeek is 69th of 123 rated AI models on Spezia's leaderboard. It costs $0.06 in, $0.18 out per million tokens.
- 69th of 123 rated models
- Rating 1,126
- Rated on 15 benchmarks
- Open weights
- Released Apr 23, 2026
- 1 million tokens of context
Where it is strongest
Not in the top tenth on any benchmark in use.
Where it falls behind
- 44th of 50 on SWE-Bench Pro
- 59th of 78 on SimpleQA Verified
- 42nd of 56 on BrowseComp
Price: $0.06 in, $0.18 out per million tokens.
Its headline results
The benchmarks that count most in the rating, with its place among the models measured the same way, and where each number came from.
- BrowseComp53.5%
Questions whose answers are hidden deep on the web and take many searches and pages to track down.
42nd of 56. The maker's own number, from DeepSeek, via LLM Stats.
- MCP Atlas67.4%
Multi-step tasks where the model has to pick and combine many real tools, connected through MCP, the standard way agents plug into apps and data.
22nd of 33. The maker's own number, from DeepSeek, via LLM Stats.
- SWE-Bench Pro52.3%
Long changes to real codebases in several languages, often across many files, including code from private and copyleft projects that models are unlikely to have seen.
44th of 50. The maker's own number, from DeepSeek, via LLM Stats.
- Humanity's Last Exam40.3%
About 2,500 questions written by experts in more than a hundred fields to be as hard as they could make them, many needing a diagram or image.
28th of 74. The maker's own number, from DeepSeek, via LLM Stats.
- Toolathlon43.5%
Long tasks across many real apps and their tools, such as email, calendars, spreadsheets, code repositories and online stores.
26th of 37. The maker's own number, from DeepSeek, via LLM Stats.
- SimpleBench61.1%
Multiple-choice puzzles about everyday situations, space, time and people, written to trip up models that pattern-match instead of think. Ordinary adults do well on them.
37th of 68. Independent result from Epoch AI.
- SimpleQA Verified33.6%
Short factual questions with one checked answer, asked without web search.
59th of 78. Independent result from Epoch AI.
- FrontierMath57.5%
Hundreds of new, unpublished math problems written by mathematicians, from hard undergraduate level up to research, with answers a computer can check.
46th of 76. Independent result from Epoch AI.
- LMArena text1,432
People put a question to two anonymous models and vote for the better answer.
61st of 107. Independent result from LMArena.
- LMArena WebDev1,582
People ask two anonymous models to build the same web page or app and vote for the better one.
33rd of 98. Independent result from LMArena.
8 more results in its full profile.
Every result with its source, what it costs per task, what jumped with its release, side by side with any other model, and answers to your questions about when to use it.
Models ranked near it
How the rating works, and every benchmark it uses: the model leaderboard.