Qwen3.5-397B-A17B: benchmarks, price and rank
Qwen3.5-397B-A17B from Alibaba is 79th of 123 rated AI models on Spezia's leaderboard, strongest on MMLU-Pro. It costs $0.45 in, $3 out per million tokens.
- 79th of 123 rated models
- Rating 983
- Rated on 12 benchmarks
- Open weights
- Released Feb 13, 2026
- 262,000 tokens of context
Where it is strongest
- 5th of 56 on MMLU-Pro
Where it falls behind
- 53rd of 56 on APEX-Agents
- 65th of 76 on FrontierMath
- 31st of 37 on Toolathlon
Price: $0.45 in, $3 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.
- APEX-Agents24.9%
Long tasks written by investment banking analysts, management consultants and corporate lawyers from their own work, done across files and apps in a realistic work setup.
53rd of 56. Independent result from Epoch AI.
- BrowseComp69%
Questions whose answers are hidden deep on the web and take many searches and pages to track down.
31st of 56. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
- Humanity's Last Exam28.7%
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.
42nd of 74. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
- Toolathlon38.3%
Long tasks across many real apps and their tools, such as email, calendars, spreadsheets, code repositories and online stores.
31st of 37. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
- LiveCodeBench v683.6%
Programming contest problems collected after models were trained, from LeetCode, AtCoder and Codeforces.
15th of 43. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
- FrontierMath31.2%
Hundreds of new, unpublished math problems written by mathematicians, from hard undergraduate level up to research, with answers a computer can check.
65th of 76. Independent result from Epoch AI.
- AA-LCR68.7%
Questions that need reasoning across several long documents at once, about 100,000 tokens of text in all.
4th of 19. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
- LMArena text1,438
People put a question to two anonymous models and vote for the better answer.
55th of 107. Independent result from LMArena.
- LMArena WebDev1,400
People ask two anonymous models to build the same web page or app and vote for the better one.
69th of 98. Independent result from LMArena.
- IMO-AnswerBench80.9%
Problems from the International Mathematical Olympiad, the hardest contest for school students, where the model has to reach the right final answer.
14th of 18. The maker's own number, from Alibaba Cloud / Qwen Team, via LLM Stats.
4 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.