Kimi K3.1 Pro beats GPT-6 Astra, nears Claude Opus 5.5

Moonshot AI released Kimi K3.1 Pro, an open-weight model with a 1M-token context and 2.8T total parameters (104B active). Terminal-Bench 4.0 scores jumped from 13% to 67% and the business-workflow test from 58% to 73%; it scores 55 on the Artificial Analysis Intelligence Index versus 58 for Claude Opus 5.5. Weights are promised by November 4.

• Terminal-Bench 4.0: 67% vs 13% for Kimi K3
• Artificial Analysis Index: 55 points, fourth place
• 1M-token context, 2.8T-parameter MoE model
• Weights promised for download by November 4