Xiaomi officially announced the MiMo-V2 series models in the early morning: The Pro model approaches Opus 4.6 with a price only 1/5 of it, and TTS supports Cantonese singing.

According to 1M AI News monitoring, Xiaomi officially released the MiMo-V2 series early this morning, covering the flagship inference model Pro, the full-modal base Omni, and the text-to-speech TTS. Yesterday, this channel reported that the model anonymously operated as "Hunter Alpha" on OpenRouter has been confirmed as MiMo-V2-Pro; this time, the three models were simultaneously announced officially and pricing was disclosed for the first time.

MiMo-V2-Pro has a total parameter count exceeding 1 trillion (42B activation parameters), supporting a 1 million Token ultra-long context, designed specifically for Agent work scenarios. It ranks eighth globally and second domestically in the Artificial Analysis leaderboard, and ranks third globally on the OpenClaw standard test benchmarks PinchBench and ClawEval. The team claims that the overall user experience surpasses Claude Sonnet 4.6 and approaches Opus 4.6, with an input pricing only 1/5 of the latter: $1 for input within 256K context, $3 per million tokens for output; $2 for input within 1M context, $6 per million tokens for output. The MiMo Claw module has integrated with the Kingsoft WebOffice ecosystem and synchronized with WPS Lingxi.

MiMo-V2-Omni is a full-modal base that supports text, image, audio, and video inputs, with a 256K context. Pricing is $0.4 for input and $2 per million tokens for output. In terms of audio, it supports over 10 hours of continuous long audio understanding, surpassing Gemini 3 Pro in comprehensive evaluation; in image understanding, it surpasses Claude Opus 4.6 and approaches Gemini 3 Pro.

MiMo-V2-TTS is based on a self-developed Audio Tokenizer, pre-trained on billions of hours of speech data and incorporating multidimensional reinforcement learning. It supports multi-granularity control from overall style to local emotion, enabling tone shifts and emotional transitions within the same sentence, and can synthesize high-quality singing voices. Dialects covered include Northeastern, Sichuan, Henan, Cantonese, and Taiwanese accents.

All three models are now integrated into Xiaomi miclaw, MiMo Studio, Kingsoft Office, and Xiaomi Browser, and can be accessed through the OpenClaw, OpenCode, KiloCode, Blackbox, and Cline five major Agent development frameworks, with a one-week free trial.