The Xiaomi MiMo team launched MiMo-V2-Pro, MiMo-V2-Omni and MiMo-V2-TTS, positioning it as a full-stack model family for the Agent era. Existing official information confirms that MiMo-V2-Pro has opened APIs, focusing on complex task orchestration, code and long context scenarios; The team's relevant statement also mentioned that there are open source plans in the future, but it needs to be promoted after the model is stable.
According to public information, MiMo-V2-Pro supports up to 1M context window, with a total parameter of more than 1T, an activation parameter of 42B, and adopts Hybrid Attention and MTP designs. The developer platform said that the first week of the launch of the new series will provide a limited-time free experience to developers around the world, but the specific quota, call limit, billing recovery time, and more complete technical details of Omni and TTS are still subject to the official follow-up instructions.
FAQs
Q: What model is MiMo-V2-Pro?
A: MiMo-V2-Pro is Xiaomi's flagship base model for Agent tasks, emphasizing long contexts, tool calls, and complex workflow processing.
Q: What are the MiMo-V2-Omni and MiMo-V2-TTS responsible for?
A: MiMo-V2-Omni focuses on multimodal understanding and action, while MiMo-V2-TTS focuses on voice generation and voice interaction capabilities.
Q: When will the new Xiaomi MiMo model be released?
A: The official page shows that the MiMo-V2-Pro page will be launched on March 18, 2026, and the Omni and TTS portals will be launched at the same time.
Q: Can developers use the MiMo API now?
A: The developer platform has opened access and announced a limited-time free trial for the first week, but subsequent billing and quota rules need to be subject to the official announcement.
