Moonshot AI Releases Kimi K3.1 Pro Model Weights, Ranks Highest Among Open-Weight Models

Huo Xing Finance reports, according to Phoenix Tech on November 5, the AI company Moonshot AI has released the model weights of Kimi K3.1 Pro on the Hugging Face platform. The AI model evaluation platform Arena revealed that Kimi K3.1 Pro (Max) ranks first among open-weight models and 13th globally among all evaluated models on the Agent Arena leaderboard for intelligent agents, securing the top position in all three categories: productivity, programming, and chat. Arena stated that this ranking is based on over 17,500 real, long-duration agent task conversations from users worldwide. Kimi K3.1 Pro (Max) achieved a net improvement rate of +5.1%, an increase of 4.2 percentage points compared to the +0.9% of its predecessor, Kimi K3 (Max). The Bash Reliability metric for shell command execution reliability reached +7.9%, ranking first among open models, compared to -4.5% and 13th place for the previous K3 model. The praise-to-complaint ratio improved to +7.7%, up from +1.3% previously; controllability reached +6.5%, also ranking first among open models. The task completion confirmation metric stood at +3.9%, placing fourth among open models, slightly higher than the previous model’s +3.8%. According to Moonshot AI, the model employs a mixture-of-experts architecture with a total of 2.8 trillion parameters, activating approximately 104 billion parameters per inference, supporting up to 1 million context tokens, and featuring native multimodal capabilities, with enhanced performance in programming, 3D content creation, financial research, factual accuracy, and complex agent task execution.