[Kimi K3 Open Sources Multiple Core Components to Support Large-Scale Intelligent Agent Training] The Kimi.ai team has gradually open-sourced several underlying infrastructures while releasing the Kimi K3 model weights and technical report. These include the high-performance communication library MoonEP for distributed MoE training, the distributed environment system AgentENV (in collaboration with kvcache-ai) designed for large-scale intelligent agent workflows, and the Kimi Delta Attention high-performance kernel FlashKDA based on CUTLASS. According to the official statement, these components can reduce communication and inference overhead in large-scale MoE and intelligent agent reinforcement learning training, and can be used as plug-and-play backends for flash-linear-attention.