2 vs 0 benchmarks won
Moonshot AI Kimi K2.5 | OpenAI GPT-5.3-Codex-Spark | |
|---|---|---|
| Overview | ||
| Company | Moonshot AI | OpenAI |
| Release date | Jan 27 2026 | Feb 12 2026 |
| Model type | — | — |
| Open source | Yes | No |
| Specifications | ||
Parameters | 1T | — |
Context window | 256k | 128k |
| Benchmarks | ||
Science reasoning GPQA Diamond | 87.6% | — |
Software engineering SWE-Bench Verified | 76.8% | — |
Multimodal understanding MMMU | — | — |
| Timeline | ||
| Release gap | Kimi K2.5 shipped 16 days before GPT-5.3-Codex-Spark | |
Kimi K2.5 leads GPT-5.3-Codex-Spark on 2 of the tracked benchmarks (GPQA Diamond, SWE-Bench Verified, MMMU). Kimi K2.5 shipped 16 days before GPT-5.3-Codex-Spark, so benchmark comparisons should account for the intervening progress.
Context windows are 256k (Kimi K2.5) vs 128k (GPT-5.3-Codex-Spark). Kimi K2.5 is an open-source / open-weight model; GPT-5.3-Codex-Spark is proprietary.
Direct benchmark comparisons are unavailable — at least one of these models has not published scores on GPQA Diamond, SWE-Bench Verified, or MMMU.