Moonshot AI's Kimi K3.1 Pro hits 55 points, fourth overall

Moonshot AI's K3.1 Pro is a post-trained version of Kimi K3, meaning the company spent significantly more compute refining the model after its initial training rather than building a larger base model. The result is a jump of 11 points on the Artificial Analysis Intelligence Index, moving from a mid-pack position to fourth place overall. The strongest gains appear in agentic tasks: on GDPval-AA the model reaches 1,846 Elo, 312 points above K3, and on Terminal-Bench 4.0 it scores 65.7 percent versus K3's 13 percent.

The model keeps the same dimensions as K3 — 2.8 trillion total parameters and 104 billion active ones — while token prices drop by 33 percent. According to Artificial Analysis, it sits at the top of the Pareto frontier, with no model of comparable parameter count scoring higher on the Intelligence Index.

Moonshot also describes deliberate limits on offensive cyber capabilities. It says vulnerability research capabilities grew as post-training scaled up, so for the open-weights release it stripped cyber-related data from both training phases. On its internal xdev-bench-hard, the uncleaned version solves 90.5 percent of tasks, while K3.1 Pro solves 81.0 percent. Moonshot says no open-weights model should push the absolute frontier of offensive cyber capabilities.