Xiaomi Open-Sources MiMo-V2.6 RL Dataset, Tops HuggingFace Trending
Xiaomi's MiMo team released the reinforcement learning dataset MiMo-V2.6-RL-oss on HuggingFace on September 26, and it quickly climbed the trending chart. Open-sourced under the Apache-2.0 license, it's regarded as a major event for open-source RL environment data.
Confirmed
- The dataset sits in the 1K–10K range (thousands of samples), spans text, image, and document modalities, and is provided in parquet format
- It contains roughly 7,000 RL environments used to train MiMo, covering code, cybersecurity, general tasks, music, and web development among other domains
- Licensed under Apache-2.0, with the community calling it a "very rare" open-source move
- Developer Harveen Chadha estimated that such RL task environments typically cost hundreds to thousands of dollars each if purchased individually, so the release carries significant commercial value
Unconfirmed
- Dorialexander, who witnessed the release, offered a key clarification: the dataset's core is 989 envs rather than the full 7,000, meaning outside interpretations of its scale are off
Why it matters
- Frontier RL training environment data is rarely open-sourced, and community members like adithyask see this as an important step forward for the field
- Comparable environment data carries a high market price, and an open version can significantly lower the barrier to RL research and training
2026-09-26 ~ 2026-09-26 · 5 related posts
Primary sources
- Xiaomi MiMo open-sources MiMo-V2.6-RL-oss training dataset on Hugging Face under Apache-2.0 — iScienceLuvr · 2026-09-26
- [source] Xiaomi Releases MiMo-V2.6-RL Dataset, Trends on Hugging Face — XiaomiMiMo · 2026-09-26
- Xiaomi open-sources ~7K RL environments used to train MiMo, spanning code, cyber and web dev — SergioPaniego · 2026-09-26
- [source] Xiaomi open-sources MiMo RL environments repo, individual tasks sell for $100-1000 each — HarveenChadha · 2026-09-26
- [source] Xiaomi open-sources ~7K RL environments, but only 989 were used to train the 9B MiMo model — teortaxesTex · 2026-09-26