MiniMax-M2.7 ranks first on Multi-SWE-bench.
MiniMax Mar 2026 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 | |
|---|---|---|---|---|
| AA-LCR (Artificial Analysis Long Context Reasoning) | 72% | #5 of 38 | Lab model card | |
| EvoArena SWE-Chain-Evo | 42.3% | #2 of 8 | Official leaderboard | |
| IFBench | 76% | #21 of 80 | Lab model card | |
| WideSearch | 75.2% | #6 of 19 | Lab model card | |
| LMArena (Elo) | 1,415 | #112 of 335 | Official leaderboard | |
| Wer wird Millionär? LLM Benchmark | 27022.0 | #27 of 77 | Official leaderboard | |
| MMLU-Pro | 81.8% | #63 of 167 | Lab model card | |
| AgentIF (2026-10 setting) | 61.9% | #13 of 24 | Official leaderboard | |
| Humanity's Last Exam | 28% | #85 of 154 | Lab model card | |
| EvoArena Terminal-Bench-Evo | 42.4% | #5 of 8 | Official leaderboard | |
| EvoArena PersonaMem-Evo | 47.9% | #6 of 8 | Official leaderboard | |
| DesignArena ASCII Art | 1,129 | #58 of 77 | Official leaderboard | |
| Extended NYT Connections | 24.7% | #86 of 103 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 12.9% | #87 of 104 | Official leaderboard | |
| HLE Extended | 6.60% | #28 of 31 | Official leaderboard | |
| BullshitBench V2 | 8.00% | #120 of 129 | Official leaderboard | |
| AA-LCR Extended | 28.4% | #46 of 49 | Official leaderboard | |
| Lech Mazur Creative Writing | -3.81 | #48 of 50 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 89.8% | #36 of 302 | Lab model card | |
| AIME 2025 | 94% | #28 of 143 | Lab model card | |
| Echo General AI Prediction Leaderboard | 1,001 | #7 of 17 | Official leaderboard | |
| AIME 2026 | 94.2% | #17 of 38 | Lab model card | |
| ProofBench | 3.00% | #55 of 59 | Official leaderboard | |
| NatureBench | 1.11% | #18 of 18 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Multi-SWE-bench | 52.7% | #1 of 10 | Lab model card | |
| LiveSQLBench | 33.2% | #6 of 36 | Official leaderboard | |
| MLE-bench | 66.6% | #2 of 10 | Lab model card | |
| VIBE-Pro (MiniMax) | 55.6% | #1 of 3 | Lab model card | |
| SWE-bench Multilingual | 76.5% | #26 of 73 | Lab model card | |
| SciCode | 47% | #18 of 49 | Lab model card | |
| DesignArena Website | 1,235 | #64 of 151 | Official leaderboard | |
| DesignArena Data Visualization | 1,230 | #51 of 118 | Official leaderboard | |
| DesignArena Game Dev | 1,199 | #65 of 143 | Official leaderboard | |
| DesignArena Code | 1,224 | #66 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,178 | #67 of 136 | Official leaderboard | |
| HyperTask | 67.6% | #1 of 2 | Lab model card | |
| DesignArena UI Component | 1,203 | #70 of 139 | Official leaderboard | |
| SWE-Bench Pro | 56.2% | #45 of 81 | Lab model card | |
| WebDev Arena | 1,398 | #67 of 105 | arena.ai, via Epoch AI | |
| DesignArena SVG | 1,124 | #68 of 103 | Official leaderboard | |
| Code Arena | 1,398 | #87 of 128 | Official leaderboard | |
| IOI (Vals) | 4.92% | #41 of 55 | Official leaderboard | |
| FrontierCode | 5.98% | #9 of 12 | Official leaderboard | |
| OpenHands Index | 43.4% | #25 of 33 | Official leaderboard | |
| NL2Repo-Bench | 39.8% | #24 of 30 | Lab model card | |
| WeirdML | 37% | #90 of 112 | Official leaderboard | |
| ALE-Bench | 1,021 | #102 of 126 | Official leaderboard | |
| Code Migration | 8.69% | #65 of 78 | Official leaderboard | |
| Vibe Code Bench | 11.9% | #101 of 120 | Official leaderboard | |
| SWE Atlas Codebase QnA Extended | 0.70% | #33 of 38 | Official leaderboard | |
| AutoLab | 0.27 | #9 of 10 | Official leaderboard | |
| SWE-Marathon | 0% | #11 of 12 | Official leaderboard | |
| EvoCode-Bench | 3.70% | #12 of 13 | Official leaderboard | |
| GameCraft-Bench | 10.9% | #14 of 15 | Official leaderboard | |
| AndroidBench | 41.6% | #30 of 32 | Official leaderboard | |
| Claw-SWE-Bench | 61.4% | #15 of 16 | Official leaderboard | |
| APEX-SWE | 21.5% | #61 of 65 | Official leaderboard | |
| SWE-bench Verified Extended | 6.70% | #44 of 44 | Official leaderboard | |
| SlopCodeBench | 4.08% | #22 of 22 | Official leaderboard | |
| EvoCode-Bench v2 | 5.10% | #13 of 13 | Official leaderboard | |
| RoadmapBench | 10.4% | #11 of 11 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench | 80.6% | #4 of 48 | Official leaderboard | |
| Toolathlon | 46.3% | #12 of 50 | Official leaderboard | |
| Claw-Eval-Live | 54.3% | #5 of 12 | Official leaderboard | |
| BrowseComp | 77.8% | #34 of 81 | Lab model card | |
| Terminal-Bench 2.0 | 45.1% | #20 of 43 | Official leaderboard | |
| RISE | 64.3% | #1 of 2 | Lab model card | |
| EnvSimBench | 41.8% | #4 of 7 | Official leaderboard | |
| SnakeBench | 22 | #111 of 188 | Official leaderboard | |
| LiveClawBench | 68.4% | #11 of 17 | Official leaderboard | |
| PinchBench 2.0 | 66.8% | #37 of 57 | Official leaderboard | |
| MM-ClawBench | 62.7% | #4 of 6 | Official leaderboard | |
| CivBench | 1,470 | #14 of 19 | Official leaderboard | |
| Clocktower Radio | 1,415 | #17 of 23 | Official leaderboard | |
| VitaBench 2.0 | 35.1% | #13 of 17 | Official leaderboard | |
| WildClawBench | 37.1% | #25 of 31 | Official leaderboard | |
| Claw-Eval | 72.6% | #20 of 24 | Official leaderboard | |
| Toolathlon-Verified | 47.5% | #21 of 25 | Official leaderboard | |
| ClawProBench | 49.5% | #42 of 49 | Official leaderboard | |
| Claw-Eval Pass^3 | 48.7% | #21 of 24 | Official leaderboard | |
| SkillsBench | 34.9% | #20 of 22 | Official leaderboard | |
| Pencil Puzzle Bench | 0% | #50 of 53 | Official leaderboard | |
| AutomationBench | 0.46% | #41 of 42 | Official leaderboard | |
| Terminal-Bench 2.1 Extended | 10.1% | #44 of 44 | Official leaderboard | |
| BrowseComp Extended | 0.50% | #42 of 42 | Official leaderboard | |
| QwenClawBench | 50.5% | #12 of 12 | Official leaderboard | |
| DeepTerminalBench | 10% | #9 of 9 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LegalBench | 84% | #44 of 138 | Official leaderboard | |
| CaseLaw v2 | 60.9% | #23 of 53 | Official leaderboard | |
| Workspace-Bench-Lite | 54.6% | #4 of 9 | Official leaderboard | |
| CorpFin v2 | 61.2% | #58 of 119 | Official leaderboard | |
| Finance Modeling Pro | 57% | #1 of 2 | Lab model card | |
| MEWC v2 | 63.3% | #1 of 2 | Lab model card | |
| GDPval-AA | 1,495 | #9 of 16 | Lab model card | |
| MedScribe | 79.9% | #56 of 97 | Official leaderboard | |
| ProfBench | 45.2% | #42 of 72 | Official leaderboard | |
| Vals Finance Agent | 48.4% | #29 of 49 | Official leaderboard | |
| Vals Index v1 | 49.9% | #24 of 34 | Official leaderboard | |
| MedCode | 34.4% | #74 of 96 | Official leaderboard | |
| TaxEval v2 | 66.6% | #102 of 129 | Official leaderboard | |
| GDPval Extended | 0.20% | #35 of 43 | Official leaderboard | |
| PhysicianBench | 8.70% | #11 of 13 | Official leaderboard | |
| Vals Index v2 | 24.6% | #57 of 65 | Official leaderboard | |
| Excel Modeling Benchmark | 28.4% | #67 of 76 | Official leaderboard | |
| Legal Research Bench | 10.6% | #83 of 93 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 0% | #72 of 80 | Official leaderboard | |
| Tax Agent Bench | 6.87% | #77 of 84 | Official leaderboard | |
| OccuBench | 53.9% | #13 of 14 | Official leaderboard | |
| Vals Finance Agent v2 | 27.9% | #75 of 80 | Official leaderboard | |
| Agents' Last Exam | 5.92% | #30 of 31 | Official leaderboard | |
| APEX-1 | 45.8% | #71 of 71 | Official leaderboard | |
| FORTE | 49.5% | #21 of 21 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MineBench | 1,194 | #57 of 75 | Official leaderboard | |
| AgentWorldBench | 46.1 | #13 of 13 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SusVibes | 18.8% | #3 of 10 | Official leaderboard | |
| nocot-bench (NCRI) | 64.0 | #123 of 185 | Official leaderboard | |
| ExploitBench | 13% | #9 of 9 | Official leaderboard | |
| SMDD-Bench | 19.3% | #7 of 7 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Artificial Analysis Intelligence Index v4.0 | 50.0 | #3 of 5 | Lab model card |
The Self-Reported Index fits one score to the 16 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 148.6 to 152.1. 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 | 57% Terminal Bench 2 (text) | 45.1% Leaderboard, IndusAGI Coding Agent | +12 pts |
| MM-ClawBench | 62.7% MM-ClawBench (MM Claw) | 62.7% Leaderboard | Same |
| Toolathlon | 46.3% Toolathlon (Toolathon in text) | 46.3% Leaderboard | Same |
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 14 of them aren't in the launch post.
