Skip to content
RobothecaBeta
Log in

Light-O1

  • AI model
  • Humanoids and robot learning

Light-O1-Preview is a model that turns a text instruction into whole-body humanoid motion, planning in language before generating the action. It is for research and robot control.

Made by LightOrigins.

Source on GitHub: github.com/lightorigins/Light-O1 (opens in a new tab). Facts read at commit 90e0568 (opens in a new tab).

Licenses

Code
Apache-2.0 (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (Apache-2.0) at commit 90e0568 (opens in a new tab), on 16 Oct 2026.

Commercial use
Yes

What it does

  • Takes a text prompt, states what the body must do in words, then generates the action.
  • Outputs a unified 138-value human action representation at 20 FPS, which can drive a character or a robot's behavior model.
  • Released as a text-to-action model only; vision and manipulation are planned for a later full version.
  • Provides a command-line tool, a Python API, and a local web console with a 3D viewer.
  • Includes a Unitree G1 simulation example in MuJoCo that needs a separate GEAR-SONIC controller checkpoint.
  • Can run on either a vLLM backend or a torch transformers backend.

Written by Robotheca from the project's README (opens in a new tab) at commit 90e0568.

Getting started

Difficulty
Advanced
Hardware it needs, as the project states it
Linux x86-64, Python 3.11, and an NVIDIA GPU compatible with CUDA 13. macOS and Windows are not supported for inference.

Maintenance

Status
Active: a commit in the last 90 days
Newest commit
8 Oct 2026 on main
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots that can run it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.