MiniMax M3: benchmarks, price and rank
MiniMax M3 from MiniMax is 66th of 123 rated AI models on Spezia's leaderboard, strongest on SWE-bench Verified. It costs $0.28 in, $1.10 out per million tokens.
- 66th of 123 rated models
- Rating 1,184
- Rated on 12 benchmarks
- Open weights
- Released May 31, 2026
- 1 million tokens of context
Where it is strongest
- 5th of 56 on SWE-bench Verified
Where it falls behind
- 13th of 13 on OfficeQA Pro
- 59th of 68 on SimpleBench
- 46th of 56 on APEX-Agents
Price: $0.28 in, $1.10 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-Agents37.7%
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.
46th of 56. Independent result from Epoch AI.
- BrowseComp83.5%
Questions whose answers are hidden deep on the web and take many searches and pages to track down.
17th of 56. The maker's own number, from MiniMax, via LLM Stats.
- MCP Atlas74.2%
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.
14th of 33. The maker's own number, from MiniMax, via LLM Stats.
- SWE-Bench Pro59%
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.
20th of 50. The maker's own number, from MiniMax, via LLM Stats.
- Vending-Bench 2$2,158
The model runs a simulated vending machine business for a year on its own: finding suppliers, ordering stock, setting prices and paying the daily fees.
49th of 60. Independent result from Epoch AI.
- OSWorld-Verified70.1%
Ordinary tasks on a real desktop computer using normal apps: a browser, a spreadsheet, an email client, a photo editor, the file manager.
17th of 25. The maker's own number, from MiniMax, via LLM Stats.
- SimpleBench45.8%
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.
59th of 68. Independent result from Epoch AI.
- MMMU-Pro78.1%
College exam questions that need a picture to answer, such as charts, diagrams, maps and chemical structures, with more answer options than the original MMMU and some questions shown only as a screenshot.
22nd of 51. The maker's own number, from MiniMax, via LLM Stats.
- LMArena text1,428
People put a question to two anonymous models and vote for the better answer.
65th of 107. Independent result from LMArena.
- LMArena WebDev1,484
People ask two anonymous models to build the same web page or app and vote for the better one.
55th of 98. Independent result from LMArena.
6 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.