Moonshot AI releases Kimi K3.1 Pro and says full open weights will follow by Nov. 4
The company says it removed cyber-related training data before the open-weight release, and its own test shows lower exploit-writing scores as a result.
As of 4:26 p.m. ET, Saturday, Oct. 24, 2026
Moonshot AI has released Kimi K3.1 Pro, an AI model with a 1-million-token context window, and says it will publish the full model weights by Nov. 4, according to the company's blog.
"Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed are available today on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. The full model weights will be released by Nov 4," the company wrote.
Moonshot describes K3.1 Pro as a post-trained version of Kimi K3 with substantially more post-training compute. It says the model improves on coding, 3D creation, financial research and other agent tasks.
Cyber data removed
The company said it found that cyber-related content in its training data was steadily raising the model's vulnerability-research ability. "For the open-weight release of K3.1 Pro, we identified and removed cyber-related data from both training stages, while retaining the general-purpose coding tasks needed for model usefulness," it wrote.
On xdev-bench-hard, Moonshot's in-house test of finding and exploiting software vulnerabilities, a version of the model trained without those measures solved 90.5% of problems, while the released K3.1 Pro solved 81.0%, the company said.
"K3.1 Pro's cyber capability remains behind that of frontier proprietary models that have been publicly available for at least three months," the company wrote.
The company said it aims not to release open weights that would extend the absolute frontier of cyber capability.
All benchmark figures are from Moonshot's own tests.
Sources
- Kimi (Moonshot AI), "Kimi K3.1 Pro Tech Blog: Advancing Open Intelligence," Oct. 24, 2026 (no time listed; read 2:49 PM ET / 18:49 UTC Oct. 24)




