Robot AI model
Octo
Octo is a generalist transformer policy conditioned on language or goal images and trained on an Open X-Embodiment mixture. The parameter count here describes the published Octo-Base checkpoint; smaller checkpoints are also available.
Records last verified Oct 13, 2026 · Send a correction
Also known as Octo-Base
- Organization
- Octo Model Team
- Type
- Policy
- Released / introduced
- May 20, 2024
- Public weights
- Public weights
- Licence / scope
- MIT
- Parameters
- 93M
- Base model
- Not disclosed
- Humanoid association
- None reported
- Inputs
- vision, language, goal image
- Outputs
- continuous action chunks
Release dates can mark an announcement or paper introduction. Public checkpoints, code licences and weight terms are distinct; follow the publisher’s links for the stated scope.
Documented embodiments
Source-associated hardware; this list does not establish plug-and-play compatibility or commercial robot capability.
- WidowX
Official repository supplies real-robot evaluation instructions.
Disclosed training data
More robot AI models
Evidence
Sources
- [s1] Octo: An Open-Source Generalist Robot Policy ↗
Octo Model Team · paper · published May 20, 2024 · accessed Oct 13, 2026
- [s2] Octo-Base model card ↗
Octo Model Team · official · accessed Oct 13, 2026
- [s3] Octo official README: checkpoints and evaluation ↗
Octo Model Team · official · accessed Oct 13, 2026