Xiaomi has released MiMo-V2.6-Pro, a flagship multimodal reasoning model based on the Mixture of Experts (MoE) architecture. The model features 1 trillion total parameters, with 42 billion active parameters used during inference. It supports text, image, speech, and video as inputs, and outputs text.
Model ReleasesXiaomiMiMo-V2.6-Pro
Xiaomi Releases MiMo-V2.6-Pro, a High-Performance Multimodal Reasoning Model
The model is designed to handle complex tasks through extended reasoning or chain-of-thought processes. It features a 1.0 million token context window and is released under the MIT license, allowing for commercial use.
In performance evaluations, MiMo-V2.6-Pro scored 46 on the Artificial Analysis Intelligence Index, significantly exceeding the median score of 18 for comparable open-weight models. It also demonstrates high efficiency, generating output at 125.2 tokens per second based on Xiaomi's API. Regarding cost, the model is priced at $0.43 per 1 million input tokens and $0.87 per 1 million output tokens.
Sources
- MiMo-v2.6-Pro: Intelligence, Performance and Price Analysis （Hacker News Frontpage, 2026-09-22）
- Hugging Face