Releases

Moonshot published Mixture-of-Block Attention code and research for efficient long-context modeling.

Moonshot introduced Kimi Linear and Kimi Delta Attention as a hybrid linear-attention approach.

Moonshot introduced the Kimina-Prover line of open models for formal theorem proving.

Moonshot described k1 as a long-context reinforcement-learning reasoning system.

Moonshot disclosed k0-math as an experimental reinforcement-learning model for mathematical reasoning.

Moonshot released Mooncake, a KV-cache-centric architecture for disaggregated large-model serving.

Moonshot began publishing open systems research for efficient long-context model training and serving.