Xiaomi-Robotics-1 (XR-1, 5B)
Robotics model
XiaomiRobotics/Xiaomi-Robotics-1-5B
Price not published
Xiaomi-Robotics-U0 (38B) / U0-4B
Xiaomi’s open embodied world model / synthetic-data engine, the generation-side companion of its VLAs (xiaomi-robotics-1).
Model idXiaomiRobotics/Xiaomi-Robotics-U0
PricePrice not published
Checked29 September 2026
Where it stands
1newer world model since its release
- Genie 427 Oct 2026
- Weights
- Weights are public under a permissive licence (Apache-2.0).
How to call it
4 ways to use Xiaomi-Robotics-U0 (38B) / U0-4B, with the model id each provider expects.
| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Hugging Face | XiaomiRobotics/Xiaomi-Robotics-U0 |
huggingface.co | |
| Hugging Face | XiaomiRobotics/Xiaomi-Robotics-U0-4B |
huggingface.co | |
| GitHub | github.com | ||
| ModelScope | modelscope.cn |
Pricing
Price not published.
No price source linked. Checked 29 September 2026.
Specs
- Input
- Text and images
- Output
- Images, video and text
- Context window
- Not published
- Max output
- Not published
- Open weights
- Yes
- Licence
- Apache-2.0
- Released
- 13 July 2026
- Status
- Current
- Type
- World model
What stands out
What its maker marketed at launch, and what people found later, each with its source.
Unified embodied synthesis
First
One autoregressive model (shared discrete visual tokenizer, next-token objective, initialized from Emu3.5) does text-to-image, image editing, multi-view robot scene generation, embodied transfer (editing scenes while keeping multi-view consistency) and embodied video rollout.
Data engine for VLAs
Synthetic data from U0 raised π0.5’s out-of-distribution success on hard real-world manipulation tasks from 36.9% to 63.2% (authors).
FlashAR fast decoding
Anti-diagonal grouped visual-token decoding plus vLLM batching: 5.44 s per 1024x1024 image on one H20, 82.86x faster than eager AR.
Notes
‘first’ flag is Xiaomi’s claim (first model with high-quality multi-view scene generation across multiple robot embodiments). Paper says 38B params; the HF README table says 34B. U0 and U0-FlashAR weights 2026-07-13; U0-4B, U0-Sequence and U0-4B-Sequence weights plus FSDP training code 2026-09-08. U0-Video announced as coming soon. Authors report beating GPT-Image-2.0 in human evals of embodied scene generation/transfer and #1 on World Arena for embodied video. Not an action model: it generates observations/data, not motor commands.
Sources: arXiv 2607.11643, HF U0-4B card, project page.