MiMo-V2.5 places in the top three on 4 benchmarks.
Open weights
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 | |
|---|---|---|---|---|
| LMArena (Elo) | 1,434 | #87 of 335 | Official leaderboard | |
| Wer wird Millionär? LLM Benchmark | 9778.0 | #35 of 77 | Official leaderboard | |
| BullshitBench | 30.9% | #49 of 107 | Official leaderboard | |
| BullshitBench V2 | 35% | #70 of 129 | Official leaderboard | |
| StudyArena Text | 1,219 | #56 of 85 | Official leaderboard | |
| DesignArena ASCII Art | 1,139 | #54 of 77 | Official leaderboard | |
| OpenAI MRCR v2 | 28% | #36 of 47 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ProofBench | 16% | #36 of 59 | Official leaderboard | |
| ResearchClawBench | 16.9 | #16 of 23 | Official leaderboard | |
| ProofBench v1.1 | 16% | #44 of 52 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DesignArena Website | 1,261 | #44 of 151 | Official leaderboard | |
| DesignArena Data Visualization | 1,250 | #41 of 118 | Official leaderboard | |
| DesignArena Game Dev | 1,243 | #50 of 143 | Official leaderboard | |
| DesignArena Code | 1,253 | #51 of 143 | Official leaderboard | |
| DesignArena UI Component | 1,250 | #52 of 139 | Official leaderboard | |
| DesignArena 3D Design | 1,200 | #59 of 136 | Official leaderboard | |
| DesignArena SVG | 1,175 | #49 of 103 | Official leaderboard | |
| WebDev Arena | 1,437 | #57 of 105 | arena.ai, via Epoch AI | |
| Code Arena | 1,438 | #72 of 128 | Official leaderboard | |
| SWE-Bench Pro | 56.1% | #46 of 81 | Lab model card | |
| Vibe Code Bench | 42.2% | #72 of 120 | Official leaderboard | |
| MiMo Coding Bench | 71.8% | #3 of 5 | Official leaderboard | |
| Code Migration | 14.2% | #57 of 78 | Official leaderboard | |
| ALE-Bench | 1,106 | #94 of 126 | Official leaderboard | |
| DesignArena Video to Website | 1,145 | #17 of 19 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench 2.0 | 89.7% | #5 of 57 | Official leaderboard | |
| Terminal-Bench 2.0 | 65.8% | #13 of 63 | Lab model card | |
| Claw-Eval | 78.4% | #6 of 24 | Official leaderboard | |
| ClawProBench | 60.4% | #21 of 49 | Official leaderboard | |
| Claw-Eval Pass^3 | 62.3% | #7 of 13 | Lab model card | |
| SnakeBench | 21 | #117 of 188 | Official leaderboard | |
| Toolathlon-Verified | 49.1% | #20 of 25 | Official leaderboard | |
| DuelLab GameBench 2 | 32.7 | #48 of 57 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CorpBench Work | 79.1 | #8 of 17 | Official leaderboard | |
| TaxEval v2 | 71.8% | #69 of 129 | Official leaderboard | |
| CorpFin v2 | 59.9% | #70 of 119 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 1.67% | #51 of 80 | Official leaderboard | |
| Excel Modeling Benchmark | 55.1% | #50 of 76 | Official leaderboard | |
| LegalBench | 78.9% | #94 of 138 | Official leaderboard | |
| Vals Index v2 | 39.9% | #45 of 65 | Official leaderboard | |
| MortgageTax | 59.3% | #63 of 87 | Official leaderboard | |
| MedScribe | 72.2% | #81 of 97 | Official leaderboard | |
| Vals Finance Agent v2 | 36.7% | #68 of 80 | Official leaderboard | |
| Tax Agent Bench | 7.77% | #73 of 84 | Official leaderboard | |
| MedCode | 31.9% | #85 of 96 | Official leaderboard | |
| Agents' Last Exam | 8.55% | #28 of 31 | Official leaderboard | |
| Legal Research Bench | 9.13% | #85 of 93 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Video-MME | 87.7% | #9 of 51 | Lab model card | |
| Claw-Eval-MM | 51.5% | #3 of 11 | Official leaderboard | |
| MMMU-Pro | 77.9% | #29 of 90 | Lab model card | |
| CharXiv | 81% | #30 of 75 | Lab model card | |
| Vision Arena | 1,234 | #58 of 136 | Official leaderboard | |
| Daily-Omni | 83.5% | #3 of 7 | Lab model card | |
| OmniCap-IF | 76.2% | #3 of 6 | Official leaderboard | |
| HR-Bench | 88.5% | #1 of 2 | Lab model card | |
| 4DCodeBench | 0.43 | #13 of 16 | Official leaderboard | |
| Video-Holmes | 64% | #5 of 6 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| nocot-bench (NCRI) | 69.9 | #102 of 185 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SAGE (Vals) | 43.3% | #50 of 81 | Official leaderboard | |
| Vals Multimodal Index | 52.8% | #23 of 33 | Official leaderboard |
The Self-Reported Index fits one score to the 6 capability results Xiaomi Corp 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 148.5 to 155.7. 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.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| Claw-Eval Pass^3 | 62.3% Claw-Eval Text | 62.3% Leaderboard | Same |
| MiMo Coding Bench | 71.8% MiMo Coding Bench | 71.8% Leaderboard | Same |
The benchmarks the lab chose to headline, with the scores it gave.