Kimi K2 (Preview): benchmark results
Provider: Moonshot. Released 2025-07-11.
Aggregate Rating 1704 ± 47, rank #307 of 10516 rated models, from 7 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| WuYuEval - Expert Module - Elo | 1991.94 | Elo rating from pairwise judged comparisons | 96.9 |
| WuYuEval - Expert Module - Judge Score | 0.85 | Mean anchor-calibrated LLM-as-a-judge score (0-1) | 87.5 |
| WuYuEval - Foundation Module - Accuracy | 92 | Accuracy (%) | 87.5 |
| WuYuEval - Foundation Module - Macro-F1 | 92.85 | Macro-F1 (%) | 84.4 |
| A.S.E (AICGSecEval) | 55.29 | Overall weighted score (0-100) | 80.6 |
| A.S.E (AICGSecEval) - Quality | 79.9 | Code quality score (%) | 77.8 |
| A.S.E (AICGSecEval) - Security | 37.82 | Code security score (%) | 69.4 |
Interactive version: theaggregate.ai/model?slug=kimi-k2-preview · How It Works · Data refreshed daily, snapshot 2026-11-21.