Moonshot AI releases Kimi K3.1 Pro model weights

What happened: Moonshot AI released the weights of its flagship multimodal model Kimi K3.1 Pro on Hugging Face, allowing independent deployment. The model was previously available through Kimi's applications and API. Key numbers: 2.8 trillion total parameters, about 104 billion activated per inference, context window of around one million tokens. Why it matters: Huawei announced inference support through vLLM Ascend and SGLang, covering its Atlas A3 and Ascend 950 systems, widening deployment options for the open weights.

Moonshot AI released the weights of its flagship multimodal model Kimi K3.1 Pro on Hugging Face, allowing independent deployment. The model was previously available through Kimi's applications and API.

Huawei announced inference support through vLLM Ascend and SGLang, covering its Atlas A3 and Ascend 950 systems, widening deployment options for the open weights.