Teams running very long agent tasks get open weights for a trillion-parameter model, though API prices could not be confirmed.
Kenning's take
What it does
MiMo-V2.5-Pro is Xiaomi's most capable model, a 1.02 trillion parameter mixture-of-experts model with 42 billion active parameters and a 1M-token context window. Xiaomi says it can keep long tasks going across more than a thousand tool calls, and shows it building a full compiler and a desktop video editor on its own. It is on Xiaomi's API platform and AI Studio, and the weights are on Hugging Face under the MIT licence.
- 1.02T total, 42B active parameters
- 1M token context window
- Long tasks over 1,000+ tool calls
- Better instruction following in agent workflows
- MIT-licensed weights on Hugging Face
Who it suits. Developers running long coding and agent workflows who want an open model.
Key facts
- Category
- AI models
- Made by
- Xiaomi
- Based in
- China
- Released
- 27 April 2026
- How to get it
- through Xiaomi's MiMo API platform (model mimo-v2.5-pro) and AI Studio, and as open weights on Hugging Face
- Works on
- API, Web, self-hosted
- Free plan
- Free MIT-licensed weights
FAQ
When was MiMo-V2.5-Pro released?
Xiaomi released and open-sourced it on 27 April 2026.
Is MiMo-V2.5-Pro open source?
Yes, Xiaomi published the weights on Hugging Face under the MIT licence.
How big is MiMo-V2.5-Pro?
1.02 trillion total parameters with 42 billion active, and a 1M-token context window.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools

ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.

Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.

Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.