MiniMax-M2.1 places in the top three on 4 benchmarks.
MiniMax Dec 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 | |
|---|---|---|---|---|
| MMLU-Pro | 88% | #5 of 167 | Lab model card | |
| LMArena (Elo) | 1,384 | #150 of 335 | Official leaderboard | |
| IFBench | 70% | #45 of 80 | Lab model card | |
| AA-LCR (Artificial Analysis Long Context Reasoning) | 62% | #22 of 38 | Lab model card | |
| Longform Creative Writing (EQ-Bench) | 57.8 | #69 of 119 | Official leaderboard | |
| Humanity's Last Exam | 22.2% | #98 of 154 | Lab model card | |
| HUMAINE | 22 | #36 of 50 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 11.8% | #77 of 104 | Official leaderboard | |
| LOCA-Bench | 5.30% | #7 of 9 | Official leaderboard | |
| Extended NYT Connections | 11.2% | #97 of 103 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 83% | #98 of 302 | Lab model card | |
| AIME 2025 | 83% | #82 of 143 | Lab model card | |
| MATH-500 | 89% | #25 of 38 | Official leaderboard | |
| SimpleBench | 34.7% | #80 of 106 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 81% | #10 of 54 | Lab model card | |
| Multi-SWE-bench | 49.4% | #3 of 10 | Lab model card | |
| LiveSQLBench | 28.7% | #12 of 36 | Official leaderboard | |
| OctoCodingbench | 26.1% | #2 of 6 | Official leaderboard | |
| SWE-bench Verified | 74% | #52 of 155 | Lab model card | |
| VIBE (MiniMax) | 88.6% | #2 of 5 | Official leaderboard | |
| DesignArena UI Component | 1,214 | #68 of 139 | Official leaderboard | |
| DesignArena Website | 1,194 | #77 of 151 | Official leaderboard | |
| SWE-bench Multilingual | 72.5% | #38 of 73 | Lab model card | |
| DesignArena Code | 1,181 | #77 of 143 | Official leaderboard | |
| DesignArena Data Visualization | 1,204 | #64 of 118 | Official leaderboard | |
| DesignArena 3D Design | 1,148 | #75 of 136 | Official leaderboard | |
| SciCode | 41% | #28 of 49 | Lab model card | |
| SWE-Bench Pro | 36.8% | #14 of 24 | Official leaderboard | |
| SWE-Review | 8.90% | #3 of 5 | Official leaderboard | |
| DesignArena SVG | 1,122 | #69 of 103 | Official leaderboard | |
| DesignArena Game Dev | 1,124 | #97 of 143 | Official leaderboard | |
| WebDev Arena | 1,387 | #72 of 105 | arena.ai, via Epoch AI | |
| FeatureBench | 41.9% | #11 of 16 | Official leaderboard | |
| FeatureBench Lite | 41.9% | #11 of 16 | Official leaderboard | |
| Code Arena | 1,384 | #93 of 128 | Official leaderboard | |
| OpenHands Index | 41.2% | #27 of 33 | Official leaderboard | |
| PostTrainBench | 9.33% | #20 of 23 | Official leaderboard | |
| CF-Div2-Stepfun | 59% | #7 of 8 | Official leaderboard | |
| ALE-Bench | 998 | #111 of 126 | Official leaderboard | |
| IOI (Vals) | 2.33% | #50 of 55 | Official leaderboard | |
| SWE-Perf | 3.10% | #1 of 1 | Lab model card | |
| SWT-bench | 69.3% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench | 80.4% | #5 of 48 | Official leaderboard | |
| tau2-bench | 87% | #22 of 79 | Lab model card | |
| Toolathlon | 40.7% | #17 of 50 | Official leaderboard | |
| Terminal-Bench 2.0 | 36.6% | #26 of 43 | Official leaderboard | |
| BrowseComp | 62% | #55 of 81 | Lab model card | |
| SnakeBench | 20 | #138 of 188 | Official leaderboard | |
| Pencil Puzzle Bench | 0% | #45 of 53 | Official leaderboard | |
| ClawProBench | 48.1% | #44 of 49 | Official leaderboard | |
| BalatroBench | 3.07 | #16 of 16 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vals Finance Agent v1 | 29.7% | #24 of 47 | Official leaderboard | |
| MedScribe | 80.8% | #51 of 97 | Official leaderboard | |
| CaseLaw v2 | 55.8% | #33 of 53 | Official leaderboard | |
| CorpFin v2 | 58.9% | #77 of 119 | Official leaderboard | |
| LegalBench | 75.5% | #106 of 138 | Official leaderboard | |
| MedCode | 34.1% | #76 of 96 | Official leaderboard | |
| TaxEval v2 | 66.3% | #103 of 129 | Official leaderboard | |
| ProfBench | 41.6% | #59 of 72 | Official leaderboard | |
| Vals Finance Agent | 33.4% | #43 of 49 | Official leaderboard | |
| GDPval-MM | 24.6% | #6 of 6 | 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 143.9 to 147.4. 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.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| Terminal-Bench 2.0 | 47.9% Terminal-bench 2.0 | 36.6% Leaderboard, Crux | +11 pts |
| Toolathlon | 43.5% Toolathlon | 40.7% Leaderboard | +2.8 pts |
| MMLU-Pro | 88% MMLU-Pro | 88% Leaderboard | Same |
| OctoCodingbench | 26.1% OctoCodingbench | 26.1% Leaderboard | Same |
| SWE-Review | 8.90% SWE-Review | 8.90% Leaderboard | Same |
| VIBE (MiniMax) | 88.6% VIBE (Average) | 88.6% Leaderboard | Same |
The benchmarks the lab chose to headline, with the scores it gave.