Moonshot · model
Kimi K2.6 benchmark scores
As of 2026-10-07, Kimi K2.6 (Moonshot) has measured scores on 50 benchmarks and estimated scores on 89 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 91.1% | measured |
| AA-HLE | 37.5% | measured |
| AA-SciCode | 51.5% | measured |
| AIME26 | 96.4% | measured |
| Artificial Analysis Intelligence Index | 27.0% | measured |
| CritPt | 8.0% | measured |
| GDPval-AA | 27.0% | measured |
| GPQA Diamond | 90.5% | measured |
| HLE | 34.7% | measured |
| IFBench | 74.6% | estimated ± 6.9 pp, medium confidence |
| AA-LCR | 81.0% | measured |
| LiveCodeBench v6 | 89.6% | measured |
| AA-Omniscience Accuracy | 32.6% | measured |
| SciCode | 52.2% | measured |
| Terminal-Bench 2.1 | 69.2% | estimated ± 6.9 pp, medium confidence |
| BrowseComp | 83.2% | measured |
| HLE w/ tools | 55.0% | estimated ± 2.4 pp, medium confidence |
| IFEval | 93.7% | estimated ± 1.3 pp, high confidence |
| LongBench v2 | 61.5% | estimated ± 5.9 pp, low confidence |
| VITA-Bench | 37.6% | estimated ± 9.9 pp, low confidence |
| τ²-bench results | 95.9% | measured |
| AA Coding Index | 61.8% | measured |
| AA-MMMU-Pro | 79.4% | measured |
| APEX-Agents | 14.6% | estimated ± 0.6 pp, low confidence |
| APEX-Agents-AA | 28.5% | measured |
| BioMysteryBench (human-difficult) | 36.7% | estimated ± 5.9 pp, low confidence |
| DeepSearchQA | 92.5% | measured |
| SWE-bench Verified | 80.2% | measured |
| AutomationBench | 28.4% | estimated ± 9.7 pp, low confidence |
| BFCL v4 | 74.5% | estimated ± 2.6 pp, high confidence |
| CyberGym | 79.0% | estimated ± 4.6 pp, high confidence |
| JobBench | 27.7% | estimated ± 4.9 pp, low confidence |
| SkillsBench | 61.3% | estimated ± 5.6 pp, low confidence |
| SWE-bench Pro | 58.6% | measured |
| τ³-bench results | 69.5% | estimated ± 1.8 pp, medium confidence |
| WideResearch | 80.8% | measured |
| AA Agentic Index | 22.1% | measured |
| MMLU-Pro | 86.5% | estimated ± 3.0 pp, high confidence |
| FrontierMath v2 (Tier 4) | 14.6% | measured |
| FrontierMath v2 (Tiers 1-3) | 39.0% | measured |
| GPQA | 90.5% | measured |
| AA-IFBench | 76.0% | measured |
| Gert Labs | 56.8% | measured |
| AA-AnalystAgent | 10.0% | estimated ± 5.5 pp, medium confidence |
| AA EnterpriseOps-Gym | 48.1% | estimated ± 3.6 pp, medium confidence |
| AA Harvey LAB | 82.0% | estimated ± 1.2 pp, medium confidence |
| ApprenticeBench | 0.0% | estimated ± 3.1 pp, low confidence |
| ARC-AGI-1 | 89.9% | estimated ± 2.7 pp, high confidence |
| ARC-AGI-2 | 57.9% | estimated ± 10.1 pp, low confidence |
| CursorBench 3.1 | 47.6% | measured |
| CursorBench 3.2 | 56.2% | estimated ± 3.6 pp, high confidence |
| Terminal-Bench 3.0 | 3.0% | estimated ± 6.9 pp, low confidence |
| FrontierCode 1.1 Main | 25.8% | estimated ± 2.7 pp, medium confidence |
| FrontierSWE v2 | 12.6% | estimated ± 9.0 pp, medium confidence |
| OfficeQA Pro | 52.6% | estimated ± 7.2 pp, medium confidence |
| OSWorld-Verified | 73.1% | measured |
| Vals GPQA Diamond | 89.1% | measured |
| Vals LiveCodeBench | 86.8% | measured |
| Vals MMLU-Pro | 87.6% | measured |
| Vals SWE-bench | 76.2% | measured |
| Terminal-Bench 2.1 (Vals AI) | 53.6% | measured |
| VulcanBench v3 | 84.9% | estimated ± 5.6 pp, low confidence |
| AA AutomationBench | 19.7% | estimated ± 5.7 pp, low confidence |
| GDP.pdf | 10.1% | estimated ± 4.9 pp, high confidence |
| AA ITBench | 27.7% | estimated ± 3.6 pp, low confidence |
| AA Tau3 Banking | 22.0% | estimated ± 5.8 pp, medium confidence |
| AA Terminal-Bench 2.1 | 66.6% | estimated ± 4.4 pp, high confidence |
| AA Terminal-Bench 4.0 | 6.5% | estimated ± 14.0 pp, low confidence |
| CursorBench 4.0 | 24.2% | estimated ± 1.6 pp, medium confidence |
| CWE-bench v1 | 37.0% | estimated ± 6.9 pp, low confidence |
| DeepSWE | 37.0% | estimated ± 6.4 pp, low confidence |
| HLE w/o tools | 33.7% | estimated ± 2.9 pp, high confidence |
| OSWorld 2.0 | 4.6% | measured |
| PostTrainBench v1.1 | 22.1% | estimated ± 2.7 pp, high confidence |
| SWE Multilingual | 76.7% | measured |
| Toolathlon-Verified | 52.5% | estimated ± 1.0 pp, low confidence |
| Vibe Code Bench | 37.9% | measured |
| CharXiv | 80.4% | measured |
| CharXiv w/o tools | 83.7% | estimated ± 1.7 pp, low confidence |
| ExploitGym | 9.8% | estimated ± 6.1 pp, low confidence |
| C-Eval | 93.1% | estimated ± 0.8 pp, low confidence |
| Claw-Eval | 62.3% | measured |
| DeepPlanning | 37.8% | estimated ± 13.1 pp, low confidence |
| HMMT Feb 2025 | 98.3% | estimated ± 1.8 pp, low confidence |
| HMMT Feb 2026 | 92.7% | measured |
| HMMT Nov 2025 | 94.4% | estimated ± 2.0 pp, high confidence |
| MathVision | 87.4% | measured |
| MCP Atlas | 55.9% | measured |
| MCP-Tasks | 76.0% | estimated ± 0.6 pp, low confidence |
| MMAnswerBench | 86.0% | measured |
| MMLU-ProX | 85.4% | estimated ± 1.7 pp, medium confidence |
| MMLU-Redux | 95.9% | estimated ± 1.1 pp, medium confidence |
| MMMU-Pro | 79.4% | measured |
| NL2Repo | 46.2% | estimated ± 3.1 pp, low confidence |
| NOVA-63 | 57.9% | estimated ± 1.6 pp, low confidence |
| QwenClawBench | 54.9% | estimated ± 2.6 pp, high confidence |
| ScreenSpot Pro | 69.7% | estimated ± 4.5 pp, medium confidence |
| SuperGPQA | 76.6% | estimated ± 7.5 pp, medium confidence |
| Terminal-Bench 2.0 | 66.7% | measured |
| Toolathlon | 50.0% | measured |
| V* | 96.9% | measured |
| VideoMMMU | 85.9% | estimated ± 1.0 pp, high confidence |
| ERQA | 65.2% | estimated ± 2.4 pp, low confidence |
| HealthBench Hard | 5.0% | estimated ± 5.1 pp, low confidence |
| LiveCodeBench Pro | 82.6% | estimated ± 6.0 pp, low confidence |
| MedXpertQA (MM) | 73.3% | estimated ± 2.4 pp, high confidence |
| MedXpertQA (Text) | 55.3% | estimated ± 4.5 pp, medium confidence |
| MMLU-Pro (Arcee) | 87.0% | estimated ± 3.4 pp, low confidence |
| React Native Evals | 75.0% | estimated ± 2.4 pp, medium confidence |
| ResearchClawBench | 18.0% | measured |
| SWE-Rebench | 61.8% | estimated ± 3.6 pp, medium confidence |
| ARC-AGI-3 | 0.0% | estimated ± 2.1 pp, medium confidence |
| FrontierMath (legacy) | 39.6% | estimated ± 0.7 pp, low confidence |
| BioMysteryBench (human-solvable) | 82.2% | estimated ± 0.6 pp, low confidence |
| DRACO | 72.6% | estimated ± 3.0 pp, medium confidence |
| FrontierCode 1.1 Extended | 24.2% | estimated ± 2.4 pp, low confidence |
| HealthBench (raw) | 46.7% | estimated ± 6.4 pp, low confidence |
| HealthBench (length-adjusted) | 53.9% | estimated ± 3.7 pp, low confidence |
| HealthBench Professional | 59.1% | estimated ± 1.3 pp, medium confidence |
| HealthBench Professional (raw) | 57.2% | estimated ± 5.0 pp, low confidence |
| HLE-Verified | 31.0% | estimated ± 2.0 pp, medium confidence |
| LABBench2 | 81.8% | estimated ± 1.8 pp, low confidence |
| AIME 2025 | 96.9% | estimated ± 1.5 pp, low confidence |
| MMMU | 86.7% | estimated ± 0.9 pp, medium confidence |
| LiveCodeBench | 100.0% | estimated ± 8.9 pp, low confidence |
| OpenHarmony Bench | 58.2% | estimated ± 2.4 pp, high confidence |
| IMOAnswerBench | 85.1% | estimated ± 5.9 pp, medium confidence |
| BabyVision | 51.6% | estimated ± 4.3 pp, low confidence |
| MMVU | 76.4% | estimated ± 2.9 pp, medium confidence |
| SimpleVQA | 66.7% | estimated ± 7.8 pp, low confidence |
| ZeroBench | 14.5% | estimated ± 4.7 pp, low confidence |
| MMMLU | 90.0% | estimated ± 1.1 pp, medium confidence |
| MMLU | 91.6% | estimated ± 1.1 pp, medium confidence |
| MMMU-Pro w/ Python | 80.1% | measured |
| MATH-500 | 97.8% | estimated ± 3.1 pp, low confidence |
| MathVision w/ Python | 93.6% | estimated ± 0.5 pp, low confidence |
| RealWorldQA | 16.4% | estimated ± 0.8 pp, low confidence |
| Video-MME (with subtitle) | 87.5% | estimated ± 0.9 pp, medium confidence |
| OmniDocBench 1.5 | 91.7% | estimated ± 5.9 pp, low confidence |