- DeepSWE v1.1 71.9
- ProgramBench
- MiMo Code Bench 63.2
- AutomationBench v1.0.6 53.1
- Toolathlon-Verified 76.9
- GDPval-AA 2.1 1673
- Agents Last Exam 31.6
- Terminal Bench 4.0 34.9
- Terminal Bench 2.1 89.9
- OSWorld-Verified 82
- JobBench 62
- CyberGym 94
- MiMo Cyber Bench 80.2
- ExploitGym 17.8
- ExploitBench 47.9
- SEC Bench Pro 66.3
- MiMo Visual Coding 72.3
MiMo-V2.6 Pro
MiMo-V2.6 Pro ranks first on MiMo Cyber Bench.
81
Benchmarks with a score
1
Ranked first, out of five or more models
5
In the top three
Xiaomi Corp Sep 2026
MiMo-V2.6 Pro
161.9
#21 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
- Terminal-Bench 4.0 27% #21 /40
- Harvey LAB (Legal Agent Benchmark) 10.8% #17 /80
- Vals Finance Agent v2 57.3% #15 /80
Where it ranks
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.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,480 | #27 of 335 | Official leaderboard | |
| StudyArena Text | 1,211 | #61 of 85 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ProofBench v1.1 | 70% | #20 of 52 | Official leaderboard | |
| Science's Last Exam | 26.4 | #6 of 9 | Official leaderboard | |
| FrontierPhysics | 1.20% | #19 of 22 | Official leaderboard | |
| MysteryMechanism | 15.3% | #29 of 31 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DesignArena Data Visualization | 1,337 | #7 of 118 | Official leaderboard | |
| DesignArena Website | 1,312 | #13 of 151 | Official leaderboard | |
| DesignArena UI Component | 1,333 | #16 of 139 | Official leaderboard | |
| DesignArena Code | 1,316 | #17 of 143 | Official leaderboard | |
| APEX-SWE | 54.2% | #8 of 65 | Official leaderboard | |
| WebDev Arena | 1,618 | #14 of 105 | arena.ai, via Epoch AI | |
| DesignArena SVG | 1,293 | #14 of 103 | Official leaderboard | |
| DesignArena 3D Design | 1,323 | #20 of 136 | Official leaderboard | |
| Vibe Code Bench | 85.2% | #18 of 120 | Official leaderboard | |
| DesignArena Game Dev | 1,313 | #22 of 143 | Official leaderboard | |
| DesignArena Frontend Web App | 1,273 | #10 of 53 | Official leaderboard | |
| Code Arena | 1,624 | #25 of 128 | Official leaderboard | |
| DesignArena Agentic Overall | 1,263 | #15 of 74 | Official leaderboard | |
| MiMo Code Bench | 63.2% | #2 of 8 | Official leaderboard | |
| Image-to-WebDev Arena | 1,597 | #16 of 55 | Official leaderboard | |
| RSI Arena | 1,025 | #3 of 10 | Official leaderboard | |
| DeepSWE | 71.9% | #17 of 56 | Lab model card | |
| Code Migration | 43% | #27 of 78 | Official leaderboard | |
| SWE-bench Verified Extended | 61.3% | #17 of 44 | Official leaderboard | |
| MiMo Visual Coding | 72.3% | #4 of 9 | Official leaderboard | |
| DAXBench v2 | 80.7% | #22 of 47 | Official leaderboard | |
| ALE-Bench | 1,476 | #60 of 126 | Official leaderboard | |
| Code Migration v1.1 | 48.8% | #17 of 30 | Official leaderboard | |
| IOI v2 (Vals) | 39.3% | #36 of 45 | Official leaderboard | |
| Tencent WorkBuddy Bench | 72.9% | #12 of 14 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Terminal-Bench 2.1 | 89.9% | #4 of 56 | Lab model card | |
| AutomationBench | 53.1% | #8 of 47 | Lab model card | |
| OSWorld-Verified | 82% | #7 of 41 | Lab model card | |
| Toolathlon-Verified | 76.9% | #6 of 21 | Lab model card | |
| DuelLab GameBench 2 | 70.9 | #18 of 57 | Official leaderboard | |
| Agent Arena | 2.40% | #19 of 56 | Official leaderboard | |
| Terminal-Bench 4.0 | 27% | #21 of 40 | Official leaderboard | |
| Terminal-Bench 4.1 | 27% | #23 of 43 | Official leaderboard |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MedScribe | 88.3% | #13 of 97 | Official leaderboard | |
| Public Benefits Bench | 68.9% | #8 of 50 | Official leaderboard | |
| Vals Finance Agent v2 | 57.3% | #15 of 80 | Official leaderboard | |
| Legal Research Bench | 47.1% | #18 of 93 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 10.8% | #17 of 80 | Official leaderboard | |
| FoodTruck Bench | $49,373 | #13 of 59 | Official leaderboard | |
| Vals Index v2 | 59.5% | #17 of 65 | Official leaderboard | |
| Tax Agent Bench | 32.4% | #23 of 84 | Official leaderboard | |
| Vals Index v2.1 | 55.2% | #15 of 49 | Official leaderboard | |
| Vals Legal Index | 72.1% | #12 of 38 | Official leaderboard | |
| Harvey LAB v1.1 (HLAB-Vals-1.1) | 13.8% | #14 of 38 | Official leaderboard | |
| Vals Finance Agent v2.1 | 64.3% | #15 of 40 | Official leaderboard | |
| MedCode | 45% | #37 of 96 | Official leaderboard | |
| Excel Modeling Benchmark | 62.9% | #30 of 76 | Official leaderboard | |
| Vals Finance Index | 57.9% | #18 of 40 | Official leaderboard | |
| APEX-Agents | 54.4% | #28 of 61 | Collected by Epoch AI, source not named | |
| Agents' Last Exam | 31.6% | #12 of 26 | Lab model card | |
| JobBench | 62% | #7 of 15 | Lab model card | |
| GDPval-AA v2.1 | 1,673 | #5 of 10 | Lab model card | |
| Immigration Bench | 67.5% | #13 of 25 | Official leaderboard | |
| APEX-Agents 1.1 | 65.3% | #35 of 65 | Official leaderboard | |
| Vals Index v2.2 | 57.8% | #20 of 37 | Official leaderboard | |
| Excel Modeling Benchmark v1.1 | 67.3% | #30 of 42 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vision Arena | 1,247 | #48 of 136 | Official leaderboard | |
| DECIDE | 40.3 | #3 of 8 | Official leaderboard | |
| Code2Video-Compass | 7.40 | #7 of 15 | Official leaderboard | |
| Humanity's Sixth Sense | 25.2% | #27 of 34 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CyberGym | 94% | #2 of 34 | Lab model card | |
| MiMo Cyber Bench | 81.7% | #1 of 5 | Official leaderboard | |
| CyberBench v1.1 (Vals) | 72.9% | #13 of 44 | Official leaderboard | |
| dfbench | 55% | #7 of 18 | Official leaderboard | |
| SAFE-Teen Urgent safety | 40% | #9 of 17 | Official leaderboard | |
| SRE-Bench | 3.05% | #21 of 35 | Official leaderboard | |
| SEC-bench Pro | 66.3% | #7 of 11 | Lab model card | |
| SAFE-Teen Relationship boundaries | 56% | #11 of 17 | Official leaderboard | |
| SAFE-Teen Health advice | 57% | #12 of 17 | Official leaderboard | |
| ExploitBench | 47.9% | #15 of 21 | Lab model card | |
| ExploitGym | 17.8% | #11 of 15 | Lab model card |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SAGE (Vals) | 45% | #41 of 81 | Official leaderboard | |
| LiveBench | 66.1% | #17 of 31 | Official leaderboard | |
| SAGE v1.1 (Vals) | 84.5% | #20 of 29 | Official leaderboard | |
| SuperCLUE | 34.6 | #16 of 16 | Official leaderboard |
Its lab's numbers put it at 161.9, number 21 of 404
The Self-Reported Index fits one score to the 8 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 158.3 to 165.6. How we compute it.
- Terminal-Bench 2.1 89.9
- OSWorld-Verified 82
- Toolathlon-Verified 76.9
- DeepSWE 71.9
- JobBench 62
- AutomationBench 53.1
- Terminal-Bench 4.0 34.9
- Agents' Last Exam 31.6
Its lab's numbers next to 4 leaderboards
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 4.0 | 34.9% Terminal Bench 4.0 | 27% Leaderboard, none, mini-swe-agent | +7.9 pts |
| MiMo Cyber Bench | 80.2% MiMo Cyber Bench | 81.7% Leaderboard | −1.5 pts |
| MiMo Code Bench | 63.2% MiMo Code Bench | 63.2% Leaderboard | Same |
| MiMo Visual Coding | 72.3% MiMo Visual Coding | 72.3% Leaderboard | Same |
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it.
MiMo-V2.6 Technical Report Sep 22, 2026
Capabilities
- DeepSWE 71.9
- ProgramBench
- MiMo Code Bench 63.2
- AutomationBench 53.1
- Toolathlon-Verified 76.9
- GDPval-AA v2.1 1673
- Agents' Last Exam 31.6
- Terminal-Bench 4.0 34.9
- Terminal-Bench 2.1 89.9
- OSWorld-Verified 82
- JobBench 62
- MiMo Visual Coding 72.3
Cyber risk
- CyberGym 94
- MiMo Cyber Bench 80.2
- ExploitGym 17.8
- ExploitBench 47.9
- SEC-bench Pro 66.3