From Startup Fortune · · 1m 20s
Moonshot AI Launches Kimi K3.1 Pro Open Weight Model
Moonshot AI's Kimi K3.1 Pro model beats Claude Opus 5.5 on coding benchmarks with free open weights by November 4.
In brief
Moonshot AI's Kimi K3.1 Pro model beats Claude Opus 5.5 on coding benchmarks with free open weights by November 4. Moonshot AI released Kimi K3.1 Pro, an open-weight AI model that scores close to Claude Opus 5.5 on coding tests and will be fully open by November 4. Originally reported by Startup Fortune.
Kimi K3.1 Pro Launch
Moonshot AI launched Kimi K3.1 Pro on October 24, its first major upgrade since the K3 release.
Coding Benchmark Results
K3.1 Pro scored 65.7% on Terminal-Bench 4.0, beating GPT-6 Astra's 58.2% and being 0.7 points behind Claude Opus 5.5.
Model Size and Features
K3.1 Pro carries 2.8 trillion parameters with 104 billion active through a mixture-of-experts architecture and handles 1 million token context.
Hallucination Rate Improvement
Kimi Hallu Bench Hard shows hallucination rate dropped from 23.4% to 8.3% between K3 and K3.1 Pro.
Integration with Tools
K3.1 Pro works directly in Blender, Unreal Engine, Godot, and CAD software for 3D modeling and game projects.
Pricing and Cost
Kimi K3.1 Pro costs $3/$15 per million input/output tokens, three to four times its predecessor's price.
Open Weights Release Date
Moonshot will publish full open weights by November 4, giving a two-week lead for monetization.
At a glance
Moonshot AI released Kimi K3.1 Pro, an open-weight AI model that scores close to Claude Opus 5.5 on coding tests and will be fully open by November 4.
