Xiaomi Livestreams MiMo-V2.6 Pro and Flash RL Post-Training Dashboard

Xiaomi's MiMo team launched a public livestream dashboard for MiMo-V2.6 post-training with parallel Pro and Flash runs. Publicly available: training steps, reward curves, token throughput, and accumulated costs; in about 1.5 days, total costs exceeded $1 million.

• Each step processes about 2 billion tokens on 1568 prompts with 16 asynchronous rollouts
• Costs: Pro ~$20,000/hour, Flash ~$10,000/hour, total over $1 million in ~1.5 days
• Pipeline fully asynchronous: generation, environment, evaluation, and weight updates run separately
• Team lead Luo Fuli promised to open technical details in the coming weeks