MiMo 7B RL 0530
The 0530 release is a trained update to MiMo 7B RL. Xiaomi expanded the supervised dataset from about 500,000 to six million examples and extended the RL training window.
The suffix identifies a changed model
The release note ties this checkpoint to a larger supervised dataset and further reinforcement learning. It is kept separate from a simple serving snapshot because its training changed.MiMo-7B-RL-0530 official model card
Compare the improvement against the original RL release
Practical guidance. Keep prompting and generation settings fixed when comparing the two models. Look for improvements across several problem types, not just the benchmark highlighted by the release. A longer training window need not help every short task.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | MiMo-7B-RL-0530 |
| Training window | Expanded from 32K to 48K during reinforcement learning; this is a training detail, not an API limit.MiMo-7B-RL-0530 official model card |
| Licence | MIT.MiMo-7B-RL-0530 official model card |
Related models
- MiMo 7B RL
Xiaomi ยท MiMo 7B
A 7B reasoning model trained with reinforcement learning for mathematics and code problems.
Launch date not confirmedDownloadable