Xiaomi has publicly opened a live dashboard for the reinforcement-learning training of its next-generation MiMo-V2.6 model, giving users access to progress, token consumption, costs, sample counts and multiple internal metrics. Luo Fuli, head of Xiaomi's MiMo team, said on September 17 that the model remains in training after the team spent nearly six months studying how far reinforcement learning can scale. The effort is expanding compute resources, training environments and task systems, and reward-evaluation mechanisms. Existing dashboard data shows MiMo-V2.6-Pro and MiMo-V2.6-Flash have spent more than $1.13 million across two days, with software-engineering tasks, data composition, failures and benchmark results also visible. Xiaomi plans to release more technical details in the coming weeks. Xiaomi-W (01810.HK) edged up 0.153%, while short-selling turnover reached approximately HK$406 million, with a short-selling ratio of 39.554%.