MiniMax-M1-80k ranks first on FullStackBench.
MiniMax Jun 2025 Open weights Flagship
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LongBench v2 | 61.5% | #10 of 35 | Lab model card | |
| OpenAI MRCR | 73.4% | #3 of 10 | Lab model card | |
| Fiction.LiveBench | 69.4% | #18 of 57 | Fiction.live leaderboard, via Epoch AI | |
| MMLU-Pro | 81.1% | #70 of 167 | Lab model card | |
| LMArena (Elo) | 1,364 | #165 of 335 | Official leaderboard | |
| SimpleQA | 18.5% | #41 of 60 | Lab model card | |
| MultiChallenge | 44.7% | #34 of 42 | Lab model card | |
| Humanity's Last Exam | 8.40% | #145 of 154 | Lab model card | |
| ConStory-Bench | 3.45 | #14 of 14 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AIME 2024 | 86% | #30 of 88 | Lab model card | |
| MATH-500 | 96.8% | #24 of 70 | Lab model card | |
| ZebraLogic | 86.8% | #11 of 23 | Lab model card | |
| GPQA Diamond | 70% | #183 of 302 | Lab model card | |
| AIME 2025 | 76.9% | #99 of 143 | Lab model card | |
| SciArena | 880 | #32 of 34 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FullStackBench | 68.3% | #1 of 8 | Lab model card | |
| SWE-bench Verified | 56% | #118 of 155 | Lab model card | |
| LiveCodeBench v6 | 65% | #69 of 90 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| tau-bench | 62.8% | #9 of 23 | Lab model card | |
| SnakeBench | 24 | #92 of 188 | Official leaderboard | |
| LisanBench | 194.7 | #91 of 133 | Official leaderboard |
The Self-Reported Index fits one score to the 15 capability results MiniMax reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 133.2 to 137.7. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it.