On March 19, 2026, Xiaomi officially launched its highly anticipated MiMo-V2 series of large-scale AI models, which includes three cutting-edge products: the MiMo-V2-Pro, MiMo-V2-Omni, and MiMo-V2-TTS. The MiMo-V2-Pro stands out as the flagship base model, boasting an impressive total of over 1 trillion parameters and the capability to support an ultra-long context of 1 million tokens. It has secured the eighth position on the global leaderboard for large-scale models and is meticulously crafted to excel in high-intensity Agent tasks. The MiMo-V2-Omni, on the other hand, is a fully multimodal base model that seamlessly integrates text, vision, voice perception, and action execution capabilities. This integration empowers it to comprehend and execute complex tasks across different modalities. Additionally, the MiMo-V2-TTS is a sophisticated large-scale speech synthesis model that offers multi-granularity control over speech styles. It can accurately replicate natural prosody and even synthesize dialects and singing voices. Notably, Xiaomi's commitment to advancing AI technology is reflected in its investment, which is set to surpass 16 billion yuan this year.