XiaomiChina2026

Xiaomi-Robotics-0

An AI model: a vision-language-action model (VLA) takes camera images and an instruction and turns them into movements for a robot. This page describes the software, not a robot.

Robotsnot stated

Key figures as stated by the developer

6 of 6 stated

Model type ⁠A vision-language-action
Controls ⁠A arms and hands
Inputs ⁠A camera language robot state
Access ⁠B open download
Parameters ⁠A 4,700,000,000
Base model ⁠A Qwen3-VL-4B-Instruct

How to read the figures

14kgA30.9lbA: The letter in the grey square after a figure points to the source the figure came from. Robots with several sources carry several letters.
74.5lb: Imperial figures are our own conversion unless the manufacturer states an imperial figure itself. The source letter always belongs to the metric figure — a conversion has no source of its own.

The four states

Versions and notes in the key figures (6)

Model type

Developer wording: "Xiaomi-Robotics-0 is an advanced Vision-Language-Action (VLA) model optimized for high performance and real-time execution.".

Controls

Xiaomi evaluates the model on "two challenging bimanual manipulation tasks, Lego Disassembly and Towel Folding".

Inputs

"The VLM processes observation images and language instructions to produce a KV cache." The action generator is "conditioned on the KV cache and the robot proprioceptive state".

Access

The authors say "code and model checkpoints are open-sourced".

Parameters

"In total, the model contains 4.7B parameters."

Base model

The vision-language part is the "pre-trained VLM ( Qwen3-VL-4B-Instruct )".

Full specifications

9 of 12 fields stated

The letter points to the source at the foot of the page. Fields without a figure are dashed: the developer does not state them — that is not the same as zero, and not the same as no.
Field	Value	Source and caveats
Input and output
Model type	`vision-language-action`	A Developer wording: "Xiaomi-Robotics-0 is an advanced Vision-Language-Action (VLA) model optimized for high performance and real-time execution.".
Inputs	`camera` `language` `robot state`	A "The VLM processes observation images and language instructions to produce a KV cache." The action generator is "conditioned on the KV cache and the robot proprioceptive state".
Controls	`arms and hands`	A Xiaomi evaluates the model on "two challenging bimanual manipulation tasks, Lego Disassembly and Towel Folding".
Degrees of freedom controlled	not stated
Size and speed
Parameters	4,700,000,000	A "In total, the model contains 4.7B parameters."
Control rate	not stated
Where it runs and what it is built on
Runs where	not stated	Note The paper says the model can "roll out fast and smoothly on real robots using a consumer-grade GPU". ⁠B
Base model	Qwen3-VL-4B-Instruct	A The vision-language part is the "pre-trained VLM ( Qwen3-VL-4B-Instruct )".
Openness
Access	open download	B The authors say "code and model checkpoints are open-sourced".
License	Apache 2.0	C The Hugging Face model card says "license: apache-2.0"; the repository names the "Apache License 2.0".
The developer’s own claims
Learns new tasks	needs retraining	A Developer claim: "Post-training adapts Xiaomi-Robotics-0 to a specific task." The earbud-packing task was learned after "post-training on 20 hours of data".
Works across robots	yes	D Developer claim: "The table below summarizes the performance results across different embodiments and datasets." The checkpoints are fine-tuned for simulation benchmarks, among them "Xiaomi-Robotics-0-SimplerEnv-Google-Robot" and "Xiaomi-Robotics-0-SimplerEnv-WidowX". Yes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count.

Notes

Xiaomi calls Xiaomi-Robotics-0 a vision-language-action model built for real-time execution; the project page is dated 12 February 2026.
Xiaomi combines a pre-trained vision-language model with a diffusion transformer that generates the actions.
Xiaomi describes the rollout technique for the robot as asynchronous execution.

Sources

A Manufacturer product page. robotics.xiaomi.com/xiaomi-robotics-0.html · Retrieved 2026-10-13
B arxiv.org/abs/2602.12684 · Retrieved 2026-10-13
C huggingface.co/XiaomiRobotics/Xiaomi-Robotics-0-Pretrain · Retrieved 2026-10-13
D github.com/XiaomiRobotics/Xiaomi-Robotics-0 · Retrieved 2026-10-13

One line per source, not per figure: a source has one retrieval date. The letter beside a figure points to the line it came from.