New modelReleasesHacker News · community
Kimi K3.1 Pro model weights are out
Moonshot AI released open weights for Kimi K3.1 Pro, a natively multimodal agentic model with a 1-million-token context window, post-trained from Kimi K3 with joint reinforcement learning. It is available via the kimi-k3.1-pro API on platform.kimi.ai with OpenAI/Anthropic-compatible endpoints, and can be self-hosted on vLLM, SGLang, or TokenSpeed. Developers must pass back the complete assistant message—including reasoning_content and tool_calls—in multi-turn conversations, and reasoning effort is set via the reasoning_effort field ("low", "high", or "max", default "max").
Read the full story on huggingface.co
What to doIf you use Kimi models, try kimi-k3.1-pro on platform.kimi.ai or download the weights, and ensure your multi-turn code returns full assistant messages including reasoning_content and tool_calls.
- Source
- huggingface.co