- Intelligence
- 23.5
- Rank 196 of 744 measured models
- Coding
- 46.8
- Rank 131 of 298 measured models
- Output speed
- 60.1 t/s
- Rank 280 of 403 measured models
- Price/1M
- $1.14
- Rank 401 of 659 measured models
- Context window
- 256K
- Rank 518 of 1199 measured models
Kimi K2.5 (Reasoning) scores 23.5 on the Artificial Analysis Intelligence Index, ranking 196th of 744 measured models. The closest models on this index are Claude Opus 4.5 (23.7), GPT-5.1 Codex (High) (23.7), and GLM 5V Turbo (Reasoning) (23.5). At $1.14 per million tokens (3:1 input/output blended price), it is cheaper than Claude Opus 4.5 ($10) and GPT-5.1 Codex (High) ($3.44). With an output speed of 60.1 t/s, it is faster than Claude Opus 4.5 (46.1 t/s), slower than GPT-5.1 Codex (High) (126 t/s).
Additional benchmarks13
Design Arena · 3d53.1%
Design Arena · Androidnative57.8%
Design Arena · Asciiart46.4%
Design Arena · Codecategories54.1%
Design Arena · Dataviz51.3%
Design Arena · Fullstack54.5%
Design Arena · Gamedev53.4%
Design Arena · Godotgamedev59.8%
Design Arena · Mobileapps54.7%
Design Arena · Svg48.4%
Design Arena · Uicomponent53.6%
Design Arena · Webapps51.6%
Design Arena · Website55.2%
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.
- MiMo-V2-Omni
Xiaomi · — · —
23.9+0.4——Compare - Claude Opus 4.5
Anthropic · $10 · 46.1 t/s
23.7+0.2$1046.1 t/sCompare - GPT-5.1 Codex (High)
OpenAI · $3.44 · 126 t/s
23.7+0.2$3.44126 t/sCompare - GLM 5V Turbo (Reasoning)
Z.AI · — · —
23.5——Compare
