2 vs 1 benchmarks won
Anthropic Claude Sonnet 4.5 | Moonshot AI Kimi K2.5 | |
|---|---|---|
| Overview | ||
| Company | Anthropic | Moonshot AI |
| Release date | Sep 29 2025 | Jan 27 2026 |
| Model type | — | — |
| Open source | No | Yes |
| Specifications | ||
Parameters | — | 1T |
Context window | — | 256k |
| Benchmarks | ||
Science reasoning GPQA Diamond | 83.4% | 87.6% |
Software engineering SWE-Bench Verified | 77.2% | 76.8% |
Multimodal understanding MMMU | 68% | — |
| Timeline | ||
| Release gap | Claude Sonnet 4.5 shipped 120 days before Kimi K2.5 | |
Claude Sonnet 4.5 leads Kimi K2.5 on 2 of the tracked benchmarks (GPQA Diamond, SWE-Bench Verified, MMMU). Claude Sonnet 4.5 shipped 120 days before Kimi K2.5, so benchmark comparisons should account for the intervening progress.
Kimi K2.5 is an open-source / open-weight model; Claude Sonnet 4.5 is proprietary.
On GPQA Diamond, Kimi K2.5 scores 87.6%, 4.2 points above Claude Sonnet 4.5 at 83.4%. On SWE-Bench Verified, Claude Sonnet 4.5 scores 77.2%, 0.4 points above Kimi K2.5 at 76.8%.