Pelican-VL
- AI model
- Humanoids and robot learning
Pelican-VL 1.0 is an open-source family of embodied multimodal brain models, from 7B to 72B parameters, that take visual and text input for robotics-oriented reasoning.
Made by Open X-Humanoid, China (checked on github.com/Open-X-Humanoid (opens in a new tab)).
Source on GitHub: github.com/Open-X-Humanoid/pelican-vl (opens in a new tab). Facts read at commit 717a1fc (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 717a1fc (opens in a new tab), on 16 Oct 2026.
- Weights
- Apache-2.0 (opens in a new tab)
Read on huggingface.co/X-Humanoid/Pelican1.0-VL-7B (opens in a new tab).
- Commercial use
- Yes
What it does
- Processes images, videos and text to recognize objects, reason about physical and spatial relations, and predict function from scene context.
- Trained on video and dynamic scene question answering to follow temporal order and object motion.
- Aims at robot tasks such as grasping, navigation and human-robot interaction, producing action sequences with per-step feasibility.
- Uses DPPO cyclic training, generating new hard samples after each reinforcement learning cycle to iteratively improve weaknesses.
- Released weights for 7B and 72B on Hugging Face and ModelScope, with a 3B checkpoint announced in December 2025.
- Includes a LoRA fine-tuning quick start built on the ms-swift framework, plus an evaluation reproduction guide.
Written by Robotheca from the project's README (opens in a new tab) at commit 717a1fc.
Getting started
- Difficulty
- Advanced
- How to start
- The project's website (opens in a new tab)
Maintenance
- Status
- Maintained: a commit in the last 12 months
- Newest commit
- 2 Dec 2025 on rxc
- Latest release
- None published on GitHub
Checked on GitHub on 16 Oct 2026.
Links
- Repository on GitHub: github.com/Open-X-Humanoid/pelican-vl (opens in a new tab)
- Pelican-VLA 0.5 (its checkpoint bundles NVIDIA's Cosmos tokenizer under the NVIDIA Open Model License): github.com/Open-X-Humanoid/Pelican-VLA05 (opens in a new tab)
- Model card: huggingface.co/X-Humanoid/Pelican1.0-VL-7B (opens in a new tab)
Robots on Robotheca that use 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.