Design Arena projects Kimi K3.1 Pro first among open-weights models; Kimi game output praised

Design Arena projects that Moonshot AI's Kimi K3.1 Pro will rank first among open-weights models on its Website leaderboard and tied for third overall, according to a post on X by the platform.

The model is 15 Elo points ahead of its predecessor, Kimi K3, and trails only Muse Wonder (max) and Muse Spark 1.3 (xhigh) across all models. The post does not include the underlying evaluation data, and the ranking is a projection, not a published leaderboard result.

A later Design Arena post says Kimi K3.1 Pro reasons more about game state, fills in details missing from prompts, and builds games that users finish. The platform says users reward these games in its arena and links to an interactive write-up on how reinforcement learning improved Kimi's game output. This is the platform's own account, and the post does not report measured results.