Xiaomi MiMo Token Plan
Global Launch
One account is eligible for one first purchase discount.
Models Intruduction
Token Plan Available
Xiaomi MiMo-V2-Pro
Foundation Model for the Agentic Era
- Trillion-parameter, efficient architecture — 1T total parameters | 42B active | 1M ultra-long context
- Ultimate Agent performance — comparable to Claude Opus 4.6 in high-intensity agent scenarios
- Hunter Alpha test volume exceeded 1T tokens, topping the OpenRouter daily chart multiple times
Token Plan Available
Xiaomi MiMo-V2-Omni
See, Hear, Act: Multimodal Agent
- Native full-modal perception — supporting image, video, audio, and text perception capabilities for cross-modal precise understanding, with perception capabilities leading the industry
- Full-modal Agent capabilities — native powerful Agent and Browser Use capabilities, the base for embodied intelligence
- Healer Alpha ranked first in average Pinchbench score during testing
Token Plan Available
Xiaomi MiMo-V2-TTS
Versatile Voice Agent that Speaks and Sings
- End-to-end architecture with proprietary Audio Tokenizer for high-fidelity speech capture
- Pre-trained on massive speech dataset for strong generalization across styles and scenarios
- Multi-dimensional RL training enabling human-like, expressive voice with unique singing and dialect capabilities