Moonshot · model
Kimi K3 (max) benchmark scores
As of 2026-10-07, Kimi K3 (max) (Moonshot) has measured scores on 15 benchmarks and estimated scores on 1 more.
| Benchmark | Score | Source |
|---|---|---|
| Terminal-Bench 2.1 | 85.0% | measured |
| Terminal-Bench 4.0 | 12.6% | measured |
| Terminal-Bench Science 0.1 | 2.6% | estimated ± 6.5 pp, medium confidence |
| GPQA Diamond | 93.5% | measured |
| SciCode | 59.5% | measured |
| CritPt | 23.4% | measured |
| GDP.pdf | 22.0% | measured |
| τ³-Bench Banking | 46.0% | measured |
| MMMU-Pro | 81.0% | measured |
| AA-AnalystAgent | 38.8% | measured |
| Harvey LAB | 94.6% | measured |
| Humanity's Last Exam | 46.9% | measured |
| AutomationBench | 58.3% | measured |
| EnterpriseOps-Gym | 45.3% | measured |
| APEX-Agents | 41.3% | measured |
| ITBench SRE | 47.7% | measured |