Xiaomi · model
MiMo-V2.6-Flash benchmark scores
As of 2026-11-21, MiMo-V2.6-Flash (Xiaomi) has measured scores on 24 benchmarks and estimated scores on 131 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 89.7% | estimated ± 9.2 pp, medium confidence |
| AA-HLE | 35.1% | measured |
| AA-SciCode | 51.3% | measured |
| Artificial Analysis Intelligence Index | 26.7% | measured |
| CritPt | 12.0% | measured |
| GDPval-AA | 55.4% | measured |
| GPQA Diamond | 89.9% | estimated ± 2.4 pp, high confidence |
| HLE | 38.2% | estimated ± 8.5 pp, medium confidence |
| AA-LCR | 74.3% | measured |
| LiveCodeBench v6 | 91.8% | estimated ± 4.9 pp, high confidence |
| AA-Omniscience Accuracy | 27.2% | measured |
| SciCode | 51.1% | estimated ± 3.1 pp, high confidence |
| Terminal-Bench 2.1 | 87.6% | measured |
| BrowseComp | 83.3% | estimated ± 3.0 pp, medium confidence |
| HLE w/ tools | 63.8% | estimated ± 4.9 pp, low confidence |
| LongBench v2 | 61.2% | estimated ± 5.5 pp, medium confidence |
| AA Coding Index | 75.5% | estimated ± 2.5 pp, high confidence |
| AA-MMMU-Pro | 73.1% | measured |
| APEX-Agents | 64.1% | estimated ± 7.4 pp, low confidence |
| APEX-Agents-AA | 41.5% | estimated ± 3.6 pp, low confidence |
| BioMysteryBench (human-difficult) | 45.1% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 86.7% | estimated ± 7.9 pp, low confidence |
| SWE-bench Verified | 87.1% | estimated ± 5.1 pp, medium confidence |
| AutomationBench | 52.3% | measured |
| BFCL v4 | 77.0% | estimated ± 2.7 pp, medium confidence |
| CyberGym | 95.1% | measured |
| JobBench | 61.2% | measured |
| SWE-bench Pro | 60.8% | estimated ± 6.3 pp, medium confidence |
| WideResearch | 83.7% | estimated ± 1.2 pp, medium confidence |
| AA Agentic Index | 51.4% | estimated ± 3.1 pp, high confidence |
| MMLU-Pro | 85.2% | estimated ± 5.0 pp, medium confidence |
| GPQA | 89.8% | estimated ± 7.8 pp, medium confidence |
| MMLU | 90.5% | estimated ± 0.9 pp, high confidence |
| Vals GPQA Diamond | 88.0% | estimated ± 5.7 pp, medium confidence |
| Vals LiveCodeBench | 84.2% | estimated ± 5.7 pp, low confidence |
| Vals MMLU-Pro | 83.8% | estimated ± 2.9 pp, high confidence |
| MMMLU | 89.8% | estimated ± 0.7 pp, medium confidence |
| MMMU | 82.2% | estimated ± 2.7 pp, high confidence |
| Chess Puzzles | 39.3% | estimated ± 8.0 pp, low confidence |
| AA MMLU-Pro | 85.2% | estimated ± 0.8 pp, high confidence |
| Gert Labs | 73.5% | estimated ± 5.5 pp, low confidence |
| AA-AnalystAgent | 41.7% | estimated ± 5.5 pp, low confidence |
| AA EnterpriseOps-Gym | 46.7% | estimated ± 1.2 pp, medium confidence |
| AA Tau3 Banking | 50.2% | estimated ± 3.3 pp, low confidence |
| AA Terminal-Bench 2.1 | 84.8% | estimated ± 2.5 pp, high confidence |
| ApprenticeBench | 35.5% | estimated ± 4.0 pp, medium confidence |
| ARC-AGI-1 | 83.7% | estimated ± 7.8 pp, low confidence |
| ARC-AGI-2 | 53.6% | estimated ± 14.4 pp, low confidence |
| CursorBench 3.1 | 47.2% | estimated ± 3.4 pp, low confidence |
| CursorBench 3.2 | 27.1% | estimated ± 3.6 pp, low confidence |
| EBR-bench | 32.3% | estimated ± 9.9 pp, medium confidence |
| Terminal-Bench 3.0 | 17.7% | estimated ± 3.2 pp, high confidence |
| FrontierCode 1.1 Main | 48.7% | estimated ± 3.5 pp, low confidence |
| FrontierSWE v2 | 21.0% | estimated ± 14.2 pp, low confidence |
| Furniture Assembly | 32.1% | estimated ± 6.5 pp, low confidence |
| MaxxBench | 56.2% | estimated ± 5.7 pp, medium confidence |
| MazeBench | 0.2% | estimated ± 3.7 pp, high confidence |
| MirrorCode | 34.5% | estimated ± 14.9 pp, low confidence |
| Mystery Game Puzzles | 12.4% | estimated ± 7.5 pp, low confidence |
| NYT Connections | 100.0% | estimated ± 8.9 pp, low confidence |
| OfficeQA Pro | 50.3% | estimated ± 6.8 pp, low confidence |
| OSWorld-Verified | 80.8% | measured |
| ParseBench-1 | 57.0% | estimated ± 6.2 pp, medium confidence |
| Terminal-Bench 4.1 | 16.1% | measured |
| Terminal-Bench Hard | 56.9% | estimated ± 2.8 pp, medium confidence |
| Vals SWE-bench | 80.2% | estimated ± 8.2 pp, medium confidence |
| Terminal-Bench 2.1 (Vals AI) | 76.4% | measured |
| CWE-Bench-AA | 61.0% | estimated ± 6.0 pp, low confidence |
| AA Cyber Index | 46.3% | estimated ± 9.4 pp, medium confidence |
| DeepsecBench-AA | 18.7% | estimated ± 5.2 pp, low confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 54.0% | estimated ± 4.3 pp, low confidence |
| HealthBench Professional (raw, Artificial Analysis) | 58.2% | estimated ± 0.8 pp, low confidence |
| AA ITBench | 43.9% | estimated ± 4.7 pp, low confidence |
| Bug Hunt Bench | 29.7% | estimated ± 10.4 pp, low confidence |
| Chartography (tools) | 79.0% | estimated ± 4.1 pp, low confidence |
| CursorBench 4.0 | 46.2% | estimated ± 4.7 pp, high confidence |
| CWE-bench v1 | 48.9% | estimated ± 4.5 pp, low confidence |
| DeepSWE | 67.9% | measured |
| DRACO | 85.4% | estimated ± 2.0 pp, medium confidence |
| FrontierCode 2.0 | 26.1% | estimated ± 9.3 pp, low confidence |
| FrontierPhysics | 0.9% | estimated ± 2.7 pp, medium confidence |
| HLE w/o tools | 37.3% | estimated ± 4.9 pp, high confidence |
| OSWorld 2.0 | 15.5% | estimated ± 3.6 pp, high confidence |
| OSWorld 2.1 | 70.4% | estimated ± 6.0 pp, medium confidence |
| OSWorld 2.1 Offline | 71.7% | estimated ± 1.9 pp, medium confidence |
| PostTrainBench v1.1 | 35.8% | estimated ± 4.5 pp, high confidence |
| ProgramBench | 26.0% | measured |
| SpireClimb | 1.4% | estimated ± 5.4 pp, low confidence |
| StealthMark | 49.3% | measured |
| Terminal-Bench 4.0 | 28.8% | measured |
| Terminal-Bench-Science 0.1 | 0.0% | estimated ± 10.7 pp, low confidence |
| Toolathlon-Verified | 73.6% | measured |
| VulcanBench Frontier v4 | 88.5% | estimated ± 1.2 pp, low confidence |
| AA AutomationBench | 40.6% | estimated ± 4.8 pp, low confidence |
| GDP.pdf | 18.7% | estimated ± 4.1 pp, medium confidence |
| AA Terminal-Bench 4.0 | 15.7% | estimated ± 6.2 pp, medium confidence |
| ARC-AGI-3 | 1.3% | estimated ± 9.8 pp, medium confidence |
| ARC-AGI-3 Provider Adapter | 2.7% | estimated ± 7.3 pp, low confidence |
| cursorBench41 | 44.8% | estimated ± 3.7 pp, medium confidence |
| CWE-bench | 45.0% | estimated ± 4.6 pp, medium confidence |
| MKA | 24.4% | estimated ± 3.1 pp, medium confidence |
| FACTS Multimodal | 34.6% | estimated ± 2.7 pp, low confidence |
| Hermes Bench | 65.7% | estimated ± 1.5 pp, low confidence |
| Hermes Index | 21.5% | estimated ± 5.5 pp, low confidence |
| ParseBench-2 | 62.0% | estimated ± 6.2 pp, medium confidence |
| CharXiv | 79.7% | estimated ± 6.3 pp, low confidence |
| CharXiv w/o tools | 79.0% | estimated ± 2.5 pp, medium confidence |
| ExploitBench | 25.3% | measured |
| ExploitGym | 6.0% | measured |
| AI2D_TEST | 89.2% | estimated ± 2.0 pp, medium confidence |
| C-Eval | 93.6% | estimated ± 1.0 pp, medium confidence |
| Claw-Eval | 76.7% | estimated ± 4.5 pp, medium confidence |
| CountBench | 93.7% | estimated ± 2.9 pp, medium confidence |
| DynaMath | 84.4% | estimated ± 3.0 pp, medium confidence |
| ERQA | 54.9% | estimated ± 3.1 pp, high confidence |
| MathVision | 82.7% | estimated ± 3.1 pp, high confidence |
| MCP Atlas | 82.2% | estimated ± 7.5 pp, medium confidence |
| MMLU-Redux | 94.4% | estimated ± 0.8 pp, medium confidence |
| MMMU-Pro | 74.4% | estimated ± 1.5 pp, high confidence |
| NL2Repo | 59.9% | estimated ± 6.0 pp, medium confidence |
| RealWorldQA | 81.8% | estimated ± 2.9 pp, medium confidence |
| ScreenSpot Pro | 73.3% | estimated ± 10.8 pp, low confidence |
| SuperGPQA | 66.6% | estimated ± 9.3 pp, medium confidence |
| Terminal-Bench 2.0 | 77.0% | estimated ± 5.8 pp, medium confidence |
| Toolathlon | 55.9% | estimated ± 6.3 pp, medium confidence |
| V* | 100.0% | estimated ± 7.5 pp, medium confidence |
| Video-MME (w/o subtitle) | 81.7% | estimated ± 1.7 pp, medium confidence |
| Video-MME (with subtitle) | 83.8% | estimated ± 3.7 pp, medium confidence |
| HealthBench Hard | 30.0% | estimated ± 8.6 pp, low confidence |
| MedXpertQA (MM) | 66.8% | estimated ± 6.5 pp, low confidence |
| React Native Evals | 77.2% | estimated ± 4.7 pp, medium confidence |
| ResearchClawBench | 19.4% | estimated ± 2.1 pp, medium confidence |
| DAYJOB | 0.7% | estimated ± 1.1 pp, low confidence |
| DAYJOB Healthcare | 1.5% | estimated ± 1.5 pp, low confidence |
| BioMysteryBench (human-solvable) | 85.1% | estimated ± 0.6 pp, low confidence |
| HealthBench (raw) | 44.1% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 51.7% | estimated ± 3.4 pp, low confidence |
| HealthBench Professional | 56.8% | estimated ± 4.2 pp, low confidence |
| HealthBench Professional (raw) | 54.9% | estimated ± 4.5 pp, medium confidence |
| HLE-Verified | 0.0% | estimated ± 6.8 pp, low confidence |
| LABBench2 | 77.7% | estimated ± 2.4 pp, low confidence |
| ScarfBench | 65.7% | estimated ± 3.8 pp, medium confidence |
| STEMBench | 2.4% | estimated ± 0.2 pp, medium confidence |
| SimpleQA | 28.0% | estimated ± 4.3 pp, medium confidence |
| Agents' Last Exam | 27.6% | measured |
| BabyVision w/ Python | 86.4% | estimated ± 4.0 pp, low confidence |
| OpenHarmony Bench v1.0 | 58.1% | estimated ± 2.4 pp, high confidence |
| SEC-Bench Pro | 47.5% | measured |
| MRCRv2 | 83.6% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 61.2% | estimated ± 2.6 pp, low confidence |
| SWE-Marathon | 42.9% | estimated ± 12.9 pp, low confidence |
| BioBasicsBench | 33.9% | estimated ± 2.8 pp, low confidence |
| MMMU-Pro w/ Python | 76.5% | estimated ± 2.0 pp, high confidence |
| ExploitBench Internal Port | 0.0% | estimated ± 6.5 pp, medium confidence |
| SWE-Atlas Codebase QnA | 60.0% | estimated ± 4.8 pp, high confidence |