3 vs 0 benchmarks won
Meta Muse Spark | Moonshot AI Kimi K2.5 | |
|---|---|---|
| Overview | ||
| Company | Meta | Moonshot AI |
| Release date | Apr 8 2026 | Jan 27 2026 |
| Model type | — | — |
| Open source | No | Yes |
| Specifications | ||
Parameters | — | 1T |
Context window | — | 256k |
| Benchmarks | ||
Science reasoning GPQA Diamond | 89.5% | 87.6% |
Software engineering SWE-Bench Verified | 77.4% | 76.8% |
Multimodal understanding MMMU | 80.4% | — |
| Timeline | ||
| Release gap | Kimi K2.5 shipped 71 days before Muse Spark | |
Muse Spark leads Kimi K2.5 on 3 of the tracked benchmarks (GPQA Diamond, SWE-Bench Verified, MMMU). Kimi K2.5 shipped 71 days before Muse Spark, so benchmark comparisons should account for the intervening progress.
Kimi K2.5 is an open-source / open-weight model; Muse Spark is proprietary.
On GPQA Diamond, Muse Spark scores 89.5%, 1.9 points above Kimi K2.5 at 87.6%. On SWE-Bench Verified, Muse Spark scores 77.4%, 0.6 points above Kimi K2.5 at 76.8%.