Moonshot AI Releases Weights for Kimi K3, a 2.8T Parameter Model

Moonshot AI has officially released the weights for Kimi K3, a massive 2.8-trillion-parameter model designed for advanced coding, reasoning, and knowledge work. This release marks the arrival of the world's first open 3T-class model with native vision capabilities and a 1-million-token context window. The release of a 2.8T-parameter open-weights model significantly pushes the boundaries of what the open-source AI community can access and run. Kimi K3's combination of a massive context window and advanced reasoning capabilities offers a powerful alternative to proprietary frontier models. Kimi K3 is built on Moonshot AI's proprietary Kimi Delta Attention and Attention Residuals architectures to manage its 1M-token context window efficiently. It supports multimodal understanding, tool calling, and parallel task execution via frameworks like Swarm and Goal.