Kimi k1.5
Multimodal long-context reasoning model at OpenAI o1 level.
About
Kimi k1.5, released by Beijing-based Moonshot AI in January 2025, is a multimodal reasoning model that matches OpenAI o1 on mathematics, coding, and multimodal reasoning. It scored 77.5 on AIME, 96.2 on MATH500, and the 94th percentile on Codeforces, with text-and-vision reasoning trained end to end.
Its technical report is notable for a deliberately simplistic RL recipe: online policy mirror descent, a length penalty against overthinking, and partial rollouts, with no MCTS, value networks, or process reward models. A long2short post-training phase compresses long chain-of-thought into short, snappy answers. Kimi k1.5 is served via API and the Kimi app, not as open weights.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 66; Trending score 70/100 from researcher assessment.