Xiaomi Releases MiMo-V2.6-Pro and MiMo-V2.6-Flash Open-Source AI Models
Xiaomi has officially released its new open-source large language models, MiMo-V2.6-Pro and MiMo-V2.6-Flash, making them available on Hugging Face. The release also includes a distilled version, MiMo-V2.6-Distill-Qwen-9B, and the models are reported to have undergone scaled reinforcement learning (RL) training. Xiaomi claims the MiMo-V2.6-Pro model achieves top performance among open-weight models, with some outlets reporting it surpasses DeepSeek R1 and scores 46 on certain benchmarks. The company also stated the training for these models involved a significant investment, estimated at over $3 million. While the availability of the models is confirmed by first-party sources, the specific performance claims and training costs are primarily reported based on Xiaomi's announcements and benchmark results.
16 articles from 10 outlets covered this story. Their coverage differs on 1 point. The underlying claim is sourced from a shipped.
What do all outlets agree on?
10 outlets covered “Xiaomi Releases MiMo-V2.6-Pro and MiMo-V2.6-Flash Open-Source AI Models”. All of them report the following:
- Xiaomi released MiMo-V2.6-Pro and MiMo-V2.6-Flash models
- The models are open-source
- They utilize reinforcement learning (RL) training
- A distilled Qwen-9B version (MiMo-V2.6-Distill-Qwen-9B) was also released
- The models are available on Hugging Face
Did outlets disagree about this?
Yes. Coverage of “Xiaomi Releases MiMo-V2.6-Pro and MiMo-V2.6-Flash Open-Source AI Models” differs on 1 point. Each account below is how a different outlet described the same event:
MiMo-V2.6-Pro is the top open weights model globally
Which figures do outlets report differently?
1 figure in this story is reported with conflicting values:
Which outlets covered this?
All 16 articles found on this story, grouped by the stance of the piece. Every link goes to the original publisher.