# Xiaomi-Robotics-1: Scaling data over model size for robot manipulation

> Xiaomi's robotics team has published a foundation model that quietly challenges one of the field's assumptions, namely that bigger networks are the fastest route to more capable robots. Xiaomi-Robotics-1 was trained on more than 100,000 hours of real manipulation data, collected with handheld grippers and cameras rather than expensive robot time, and the payoff shows up on the benchmarks. It scored 57.4 on RoboCasa365 against 46.6 for the next best system, and success rates on hard tasks climbed from roughly 25 percent to 75 percent as the training data grew. The interesting claim is not the leaderboard position but the recipe behind it. Where much of the industry has chased scale in parameters, this work argues that scale in demonstrations matters more for teaching robots to move, grasp, and adjust in the physical world. Because the model is hardware agnostic, it is meant to drop into different robot fleets rather than a single custom platform. If the result holds up beyond Xiaomi's own tests, it points to cheaper data collection, not ever larger models, as the practical path to useful manipulation, which would reshape how startups and labs budget their robotics efforts.

_Section: [Daily AI Updates](https://www.wortins.com/daily-ai) · Source: The Decoder · Published Thursday, July 23, 2026_

## Wortins' read

Xiaomi's robotics team has published a foundation model that quietly challenges one of the field's assumptions, namely that bigger networks are the fastest route to more capable robots. Xiaomi-Robotics-1 was trained on more than 100,000 hours of real manipulation data, collected with handheld grippers and cameras rather than expensive robot time, and the payoff shows up on the benchmarks. It scored 57.4 on RoboCasa365 against 46.6 for the next best system, and success rates on hard tasks climbed from roughly 25 percent to 75 percent as the training data grew. The interesting claim is not the leaderboard position but the recipe behind it. Where much of the industry has chased scale in parameters, this work argues that scale in demonstrations matters more for teaching robots to move, grasp, and adjust in the physical world. Because the model is hardware agnostic, it is meant to drop into different robot fleets rather than a single custom platform. If the result holds up beyond Xiaomi's own tests, it points to cheaper data collection, not ever larger models, as the practical path to useful manipulation, which would reshape how startups and labs budget their robotics efforts.

## Source

[Read the full story at The Decoder](https://the-decoder.com/xiaomi-robotics-1-shows-that-more-data-beats-bigger-models-when-training-robots-to-move/)

## Related coverage

- [Clearview AI Is Testing an AI Tool That Would Let Cops Unearth Your Life Online](https://www.wortins.com/story/clearview-ai-is-testing-an-ai-tool-that-would-let-cops-unear-8c573d0e) — [Wired](https://www.wired.com/story/clearview-ai-is-testing-an-ai-tool-that-lets-cops-instantly-unearth-your-online-activity/)
- [Powering AI is an architecture problem](https://www.wortins.com/story/powering-ai-is-an-architecture-problem-15281f16) — [MIT Technology Review](https://www.technologyreview.com/2026/09/10/1141649/powering-ai-is-an-architecture-problem/)
- [Inception launches Mercury 2.5 at 1,107 tokens per second](https://www.wortins.com/story/inception-launches-mercury-2-5-at-1-107-tokens-per-second-7999ed1b) — [TestingCatalog](https://www.testingcatalog.com/inception-launches-mercury-2-5-at-1-107-tokens-per-second/)
- [STAT+: Can AI fix the emergency room?](https://www.wortins.com/story/stat-can-ai-fix-the-emergency-room-a71d00ee) — [STAT](https://www.statnews.com/2026/09/09/scribe-emergency-room-fix-health-care-ai-prognosis/?utm_campaign=rss)
- [Top AI spenders cut per-employee costs by nearly 10 percent in August](https://www.wortins.com/story/top-ai-spenders-cut-per-employee-costs-by-nearly-10-percent--b31ffb0b) — [The Decoder](https://the-decoder.com/top-ai-spenders-cut-per-employee-costs-by-nearly-10-percent-in-august/)
- [Researchers used AI to build a WeChat worm that spreads through phone calls](https://www.wortins.com/story/researchers-used-ai-to-build-a-wechat-worm-that-spreads-thro-8bf3f469) — [The Next Web](https://thenextweb.com/news/wechat-worm-ai-calif-tencent-zero-click)

---

_Curated and written by [Wortins](https://www.wortins.com) — The daily AI briefing. Every story links to its original source; the "Wortins read" on each is our own original analysis. [About Wortins & our editorial approach](https://www.wortins.com/about)._