Kimi K3 Open Model Scores Above GPT-5.6 Sol in Tests
Kimi K3 matches or exceeds GPT-5.6 Sol and Claude Fable 5 on several benchmarks while remaining open-source and cost-effective.
Reporting from 1 source: GIGAZINE.
Chinese AI company Moonshot AI announced Kimi K3, a 2.8 trillion parameter open model with a 1 million token context window. The model scored higher than GPT-5.6 Sol and Claude Fable 5 on several benchmarks, including coding and agent performance. Moonshot plans to release the full model as open source on July 27, 2026. API pricing is $3 per million input tokens and $15 per million output tokens.
Moonshot AI released Kimi K3 on July 17, 2026, with a parameter count of 2.8 trillion and a context window of 1 million tokens. In third-party evaluations by Artificial Analysis, Kimi K3 surpassed Claude Opus 4.8 and GPT-5.6 Terra on the Intelligence Index, as well as on coding and agent performance indexes. The model is available now on Kimi.com and Kimi Code, with API pricing starting at $3 per million tokens for input. The full weights will be published as open source on July 27.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.