Kimi K3 (Max) — benchmark results
Provider: Moonshot. Released 2026-07-17. Access: API.
Unified ELO 2013 ± 57, rank #24 of 1756 rated models, from 8 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLM2014 Logic 2026-07 | 77.04 | Median Score | 97.6 |
| VoxelBench | 2085 | Rating | 95.7 |
| OTIS Mock AIME 2024-25 | 97.22 | Accuracy (%) | 93.6 |
| SimpleBench | 60.7 | Score (AVG@5) | 77.3 |
| Chess Puzzles (Epoch AI) | 39 | Accuracy (%) | 75 |
| FrontierMath - Tiers 1-3 (v2) | 72.18 | Accuracy (%, 285 private v2 problems) | 73 |
| FrontierMath - Tier 4 (v2) | 39.02 | Accuracy (%, 41 private v2 problems) | 69.2 |
| SimpleQA Verified | 42.7 | Accuracy (%) | 57.8 |
Interactive version: aibenchmarks.dev/model?slug=kimi-k3-max · How the rankings work · Data refreshed daily, snapshot 2026-07-18.