Moonshot AI has released the weights of Kimi K3.1 Pro on Hugging Face, making its flagship multimodal model available for independent deployment. The model uses a mixture-of-experts architecture with 2.8 trillion total parameters and approximately 104 billion activated per inference, and supports a context window of around one million tokens.
K3.1 Pro builds on Kimi K3 with expanded reinforcement learning for coding, 3D creation, financial research and long-running agent tasks. The weights release follows the model’s earlier availability through Kimi’s applications and API. Huawei has also announced inference support through vLLM Ascend and SGLang, covering its Atlas A3 and Ascend 950 systems. [Moonshot AI]