Xiaomi open-sourced MiMo-V2.6 Pro and Flash. Both are omnimodal models for coding, visual tasks and computer use. Pro is the most capable model. Flash balances intelligence and efficiency. Pro-UltraSpeed offers up to 20 times faster output at the same quality. Pro scored 46.32 on the Artificial Analysis Intelligence Index v4.3. Xiaomi says this puts Pro ahead of Kimi K3 and Qwen3.8 Max. API prices remain at V2.5 levels. UltraSpeed costs ten times more than standard output. Models are available now in Xiaomi’s platforms and OpenRouter. MiMo Desktop is leaving early access. Xiaomi published the technical report, training environments and RL code. The launch includes reproducible tests of scaled reinforcement learning.
What changed
MiMo-V2.5 models were the latest open-sourced series.
What it unlocks
Running omnimodal models for coding, visual tasks, computer use and real-time workflows.
- 46.32 on Artificial Analysis Intelligence Index v4.3
- $0.14 per million input tokens for Flash
- $0.435 per million input tokens for Pro
- 20 times faster output for Pro-UltraSpeed
What you need to act on it
- access to AI Studio, MiMo Code, MiMo Desktop, Xiaomi’s MiMo API Platform or OpenRouter
Sources