0 vs 2 benchmarks won
Anthropic Claude Sonnet 4 | Moonshot AI Kimi K2.5 | |
|---|---|---|
| Overview | ||
| Company | Anthropic | Moonshot AI |
| Release date | May 22 2025 | Jan 27 2026 |
| Model type | — | — |
| Open source | No | Yes |
| Specifications | ||
Parameters | — | 1T |
Context window | — | 256k |
| Benchmarks | ||
Science reasoning GPQA Diamond | 75.4% | 87.6% |
Software engineering SWE-Bench Verified | 72.7% | 76.8% |
Multimodal understanding MMMU | — | — |
| Timeline | ||
| Release gap | Claude Sonnet 4 shipped 250 days before Kimi K2.5 | |
Kimi K2.5 leads Claude Sonnet 4 on 2 of the tracked benchmarks (GPQA Diamond, SWE-Bench Verified, MMMU). Claude Sonnet 4 shipped 250 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 is proprietary.
On GPQA Diamond, Kimi K2.5 scores 87.6%, 12.2 points above Claude Sonnet 4 at 75.4%. On SWE-Bench Verified, Kimi K2.5 scores 76.8%, 4.1 points above Claude Sonnet 4 at 72.7%.