ModelFeatured
Octo
Generalist robot policy Transformer with multi-modal inputs and multi-embodiment finetuning.
93MFully openData: Open X-EmbodimentUpdated 2024-07-31
ModelFully open
Uni-O4 / Offline RL Policies
Open offline RL and unified policy learning implementations for data-driven control research.
Overview
Uni-O4 represents research toward unified world-model / policy objectives for embodied agents. Treat it as a paper+code entry for world-model-centric learning rather than a batteries-included product framework.
Who it is for
World-model researchers; model-based RL for robots.
Key highlights
- World-model research direction
- Unified objective narrative
- Open research code (verify repo)
When to use
- Exploring world models for control
When not to use
- Need a stable IL product baseline this week
Getting started
- 1Read the paper
- 2Reproduce toy experiments
- 3Port ideas into your stack carefully
How it compares
Caveats & pitfalls
- Less community packaging than LeRobot/OpenVLA.
Content reviewed 2026-07-23