Moonshot AI opens Kimi K3.1 Pro weights on Hugging Face

Moonshot AI opens Kimi K3.1 Pro weights on Hugging Face

N
News Editor
2026-11-05 02:34:06
Moonshot AI has released the weights for Kimi K3.1 Pro on Hugging Face, according to a Nov. 5 report cited by ChainCatcher from Phoenix Technology. Arena, an AI model evaluation platform, said Kimi K3.1 Pro (Max) ranked first among open-weight models on the Agent Arena leaderboard and 13th among all models evaluated globally. It also placed first among open models across work, coding, and chat categories. Arena said the ranking was based on more than 17,500 real, long-horizon agent task conversations from users worldwide. On performance metrics, Kimi K3.1 Pro (Max) posted a net improvement rate of +5.1%, up 4.2 percentage points from the previous-generation Kimi K3 (Max) at +0.9%. Its Bash Reliability score came in at +7.9%, ranking first among open models, versus -4.5% and 13th place for the prior K3. Moonshot AI said the model uses a mixture-of-experts architecture with 2.8 trillion total parameters and about 104 billion activated parameters per inference. It supports roughly 1 million tokens of context, includes native multimodal capabilities, and is designed to improve coding, 3D creation, financial research, factual accuracy, and execution of complex agent tasks.

Moonshot AI has opened the weights for Kimi K3.1 Pro on Hugging Face, according to a Nov. 5 report from Phoenix Technology cited by ChainCatcher.

Arena, an AI model evaluation platform, said Kimi K3.1 Pro (Max) ranked No. 1 among open-weight models on the Agent Arena leaderboard and No. 13 among all models evaluated globally. It also placed first among open models in work, coding, and chat.

Arena said the ranking was based on more than 17,500 real, long-horizon agent task conversations from users around the world.

Kimi K3.1 Pro (Max) posted a net improvement rate of +5.1%, up 4.2 percentage points from +0.9% for the previous-generation Kimi K3 (Max). Its Bash Reliability score reached +7.9%, ranking first among open models. The previous K3 recorded -4.5% and ranked 13th.

The model’s praise-and-complaint metric came in at +7.7%, compared with +1.3% for the previous generation. Its controllability metric was +6.5%, also ranking first among open models. On task completion confirmation, it scored +3.9%, ranking fourth among open models and slightly above the previous generation’s +3.8%.

According to Moonshot AI, the model uses a mixture-of-experts architecture with 2.8 trillion total parameters and about 104 billion activated parameters per inference. It supports roughly 1 million tokens of context, has native multimodal capabilities, and is aimed at improving coding, 3D creation, financial research, factual accuracy, and execution in complex agent tasks.

This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan.
4600

Disclaimer:

The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.

Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.

Moonshot AI opens Kimi K3.1 Pro weights on Hugging Face | Bit.Fan