Moonshot AI's Kimi K3.1 Pro Tops Artificial Analysis Index at 55
A Chinese model that posts a higher score than the flagship models of leading American AI companies on an independent index, and then gives itself away free, puts the pricing side of the frontier race under pressure as much as the benchmark side.
Reporting from 1 source: GIGAZINE.
Moonshot AI announced Kimi K3.1 Pro on October 24, 2026, a 2.8-trillion-parameter MoE model with 104 billion active parameters and a 1-million-token context window. On the Artificial Analysis Intelligence Index v4.3.2 it scored 55, above Claude Fable 5.1, GPT-6 Astra, and Gemini 4 Argon at 53. The model arrives free of charge on November 4, 2026, with API pricing at $2.00 input and $10.00 output per million tokens.
Kimi K3.1 Pro is a mixture-of-experts model, and the parameter split matters: 2.8 trillion total, 104 billion active. Moonshot pitches it at coding agents and agent tasks, and it also drives 3D applications including Blender and FreeCAD. On Terminal-Bench, which scores practical terminal work against cost per task, the model recorded a higher score at lower cost than GPT-6 Astra and Claude Fable 5.1. Access runs through Kimi.ai, Kimi Work, Kimi Code, and the API. A Highspeed version triples speed at twice the price.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.