Moonshot AI releases Kimi K3.1 Pro with 2.8 trillion parameters

Moonshot AI has released the Kimi K3.1 Pro model, a post-trained version of its predecessor, Kimi K3. The new model features a parameter count of 2.8 trillion and supports both text and image input. It is available through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.

Kimi K3.1 Pro scored 55 points on the Artificial Analysis Intelligence Index, placing it fourth on the ranking. The model is positioned on the Pareto frontier according to Artificial Analysis. It generates output at a rate of 77 tokens per second, with a time to first token of 2.89 seconds. The model has a context window of one million tokens.

API input costs for the model are $2 per million tokens, while output costs are $10 per million tokens. These prices represent a 33 percent drop compared to the previous version. On the GDPval-AA benchmark, the model reached 1,846 Elo. It scored 92.7 percent on the BenchCAD benchmark with tools. Internal testing on the Kimi Hallu Bench Hard showed a reduction in claim-level hallucinations from 23.4 percent to 8.3 percent.

Moonshot AI plans to release the model weights by November 2026.