Modelmultimodalv7B

OpenVLA

by Stanford University (Academic) · open-source · Last verified 2026-03-17

OpenVLA is Stanford University's open-source vision-language-action model for robot manipulation, built on a 7B parameter vision-language model backbone and fine-tuned on the Open X-Embodiment dataset. It achieves strong performance on robot manipulation tasks while being fully open-source and reproducible, serving as a community standard for VLA research.

https://openvla.github.io/ ↗

C+

C+—Average

Adoption: CQuality: AFreshness: B+Citations: BEngagement: F

Specifications

License: MIT
Pricing: open-source
Capabilities: vision-language-action, robot-manipulation, language-conditioned-control, multi-task-robot-policy
Integrations: Hugging Face, PyTorch, ROS 2
Use Cases: robot manipulation research, language-conditioned robot control, VLA model fine-tuning, academic robot learning
API Available: Yes
Parameters: ~7B
Context Window: N/A
Modalities: vision, text, action
Training Cutoff: 2024
Tags: robotics, open-source, vision-language-action, stanford, manipulation
Added: 2026-03-17
Completeness: 100%

Index Score

51.5

Adoption

Quality

Freshness

Citations

Engagement

Explore the full AI ecosystem on Agents as a Service