LLM
MiMo-V2.6-Pro-RL
MiMo-V2.6-Pro-RLXiaomi’s flagship omnimodal MoE model, with 1.02T parameters, 42B active, a 1M-token context window, and RL training for coding and agents.
Model details
View repositoryMiMo-V2.6-Pro-RL is Xiaomi’s flagship omnimodal AI model, built on a sparse Mixture-of-Experts architecture with 1.02T total parameters (42B activated) and a hybrid sliding-window/global attention design supporting context windows up to 1M tokens. It natively processes text, images, video, and audio through dedicated vision and audio encoders while producing text outputs. Trained through large-scale mixed reinforcement learning and agentic grading, it excels at multimodal perception, long-context reasoning, coding, cybersecurity, and agentic workflows involving multi-step tool use.
