MiMo-V2.6-Pro-UltraSpeed

MiMo-V2.6-Pro-UltraSpeed is the high-speed serving tier the Xiaomi MiMo platform offers for MiMo-V2.6-Pro. Xiaomi says it keeps V2.6-Pro's capability while delivering up to 20x output speed, aimed at real-time interaction and latency-sensitive production workloads. Xiaomi has not documented how the tier differs technically from the standard tier, and has not published separate benchmark results for it.

Specifications

Capabilities match MiMo-V2.6-Pro: a sparse MoE with 1.02 trillion total parameters and 42B activated. The models are natively omni-modal: they accept text, image, video and audio input and return text, with a 1M-token context window. The vision encoder is a 681M-parameter MiMo ViT (28 layers, 24 SWA + 4 full attention); the audio stack combines a 308M AudioTokenizer with a 127M audio patch encoder, and a 5-layer MTP speculative decoder predicts 7 tokens per pass.

Pricing

The tier is priced separately: ¥30 per million input tokens, ¥0.25 per million cached input tokens and ¥60 per million output tokens domestically; internationally $4.35 input, $0.036 cached input and $8.7 output per million tokens — roughly 10x the standard tier.

Release and license

The MiMo-V2.6 series was published on Hugging Face on September 21, 2026 under the MIT license, and served through the Xiaomi MiMo platform (mimo.mi.com) from September 22, with distribution also via ModelScope and OpenRouter. Xiaomi published a live dashboard of the RL post-training run: Pro and Flash each took under six days and 30 steps, about 750,000 trajectories in total, at a reported cost of roughly $2.62M and $0.85M respectively.