Add as a preferred source on GoogleMoonshot AI Publishes Kimi K3.1 Pro Weights on Hugging Face
The release adds published weights for a model already available through Moonshot AI’s API. Kimi K3.1 Pro has 2.8 trillion total parameters and a context window of about 1 million tokens.
Moonshot AI has published Kimi K3.1 Pro’s weights on Hugging Face, giving developers access to the model files for a system that was already available through its API.
The API supports interfaces compatible with OpenAI and Anthropic. Developers can call the model by selecting kimi-k3.1-pro on platform.kimi.ai.
Kimi K3.1 Pro is the first model in the K3.1 series and uses a mixture-of-experts architecture. It has 2.8 trillion total parameters, including 104 billion active parameters, and a context window of about 1 million tokens.
The weights use MXFP4 quantization, while activations use MXFP8. Training used quantization-aware methods.
Moonshot AI set Nov. 4 as the target for publishing the weights in its Oct. 23 launch announcement. The company’s weight-release plan was part of that announcement.
Simon Yoon
ReporterSimon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.