Xiaomi launches MiMo-V2.6-Pro amid Anthropic data-use allegation

Xiaomi released its MiMo-V2.6 lineup, led by MiMo-V2.6-Pro and accompanied by MiMo-V2.6-Flash. The company said Pro scored 46 on Artificial Analysis’s Intelligence Index, ahead of rivals including Kimi K3 and Qwen, and said its pricing was $0.435 per million input tokens and $0.87 per million output tokens.

Xiaomi said the improvement came from a larger reinforcement-learning run that scaled data, environments and compute for grading. It said the run took less than six days and cost about $2.62 million for Pro and $0.85 million for Flash; the article also said Artificial Analysis put a single test task at about $0.13.

The company also said it was releasing Pro-UltraSpeed, a faster variant, and opening up its RL toolkit. That package included the technical report, the training framework, a smaller model for further training, around 7,000 tasks with automatic graders, and about 1,000 music composition tasks.

Anthropic had already published a threat intelligence report naming Xiaomi and six other Chinese labs, including Alibaba, Moonshot AI, DeepSeek, Zhipu, MiniMax and SenseTime. In the Xiaomi case, tagged GTG-16008, Anthropic said it tracked more than 400,000 exchanges over 20 days in March and April 2026, and said Xiaomi passed user conversations and coding sessions from its MiMo models through OpenClaw and OpenCode to Claude for training data.