Xiaomi is the latest company to release an open-weight AI model – MiMo-V2.5 claims it is “a major step forward in agentic capability and multimodal understanding.”
Xiaomi has shared various benchmark results that compare the MiMo-V2.5 to the recently released DeepSeek-V4, Kimi K2.6, Cloud Opus 4.6, Gemini 3.1 Pro, and Xiaomi’s older MiMo-V2-Pro.
The company claims that MiMo-V2.5 has achieved best-in-class performance on its in-house agentive task benchmark. On the internal MiMo coding bench, the smaller V2.5 model matches the larger V2.5-Pro at half the price. Xiaomi says that in other benchmarks testing the model’s image and video understanding, V2.5 is at the level of closed-source models.
MiMo-V2.5 was evaluated on coding and agentic tasks.
The model was trained on 48 trillion tokens and is natively multimodal with support for text, image, and video data. Xiaomi has published two versions: MiMo-V2.5 with 310B total parameters (15B active) and MiMo-V2.5-Pro with 1.02T total parameters (42B active). The model supports up to 1 million tokens of reference.
MiMo-V2.5 evaluated on image and video understanding
You can download the model from here hugging face And run it yourself, but you’ll need something like a kitted-out Mac Studio to do it – consumer GPUs don’t have enough VRAM (no, not even the Nvidia RTX 5090).
You can try Xiaomi MiMo-V2.5 AI Studio (which is not loaded at the time of writing) or use this via official api. Or, as mentioned above, download it and run it locally, if you have the means to do so.
