Kimi K3.1 Pro launches at fourth on Artificial Analysis Intelligence Index

What happened

Nathan Lambert notes that Kimi K3.1 Pro scores higher than Fable 5.1 and Astra 6. Quoting Artificial Analysis, the model ranks fourth on its Intelligence Index with a score of 55, gaining 11 points over Kimi K3 and making large gains in agentic performance. Artificial Analysis says it will lead open-weights models once its weights are published.

Read next

• More changes in today’s briefingOlmoEarth v1.3 releases global satellite embeddings and a browser-based mapping explorer

Reported by 3 sources

The story so far

Moonshot AI released Kimi K3.1 Pro, which Artificial Analysis ranked fourth on its Intelligence Index with a score of 55—11 points above Kimi K3 and three points behind Claude Opus 5.5. Artificial Analysis said the reasoning model made large gains in agentic performance, while Nathan Lambert noted that its score exceeded those of Fable 5.1 and Astra 6. The model is available through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. Moonshot also offers a Highspeed variant that it says runs two to three times faster. The Decoder reported a 33 percent reduction in token prices. Artificial Analysis said Kimi K3.1 Pro would lead open-weights models once its weights are published. The Decoder reported that Moonshot plans to release the weights by November 4, 2026; that lead remains conditional on the release and rankings remaining unchanged.