Model comparison · robotics
OpenVLA 7B vs pi 0.7
Symmetrical, primary-source facts with unknown values left visible. No winner is assigned without comparable measurements.
FieldAt a glance
OpenVLA Research Team · activeOpenVLA 7BVerified Aug 28, 2026
Physical Intelligence · activepi 0.7Verified Aug 28, 2026
Technical differences
Side-by-side facts
| Field | OpenVLA 7B | pi 0.7 |
|---|---|---|
| Developer | OpenVLA Research Team | Physical Intelligence |
| Family | OpenVLA | pi |
| Model | OpenVLA 7B | pi 0.7 |
| Version | 7B | 0.7 |
| Lifecycle | active | active |
| Input modalities | Text, Image, Robot state | Text, Image, Robot state |
| Output modalities | Robot action | Robot action |
| Context window | Unknown | Unknown |
| Total parameters | 7000000000 | Unknown |
| Active parameters | Unknown | Unknown |
| License | Unknown | Unknown |
| Open weights | Yes | No |
| API available | No | Unknown |
| Self-hostable | Yes | Unknown |
| Capabilities | cross-embodiment, fine-tuning, generalist-manipulation | cross-embodiment, dexterous-manipulation, language-steering, visual-subgoals |
| Robotics model type | Vision-language-action model | Vision-language-action model |
| Action representation | Tokenized actions decoded to continuous robot controls | Continuous robot actions conditioned by multimodal prompts |
| Control architecture | Fused SigLIP and DINOv2 visual encoder with Llama 2 7B backbone | High-level policy, world model, and action expert |
| Inference location | Flexible | Unknown |
| Native control rate (Hz) | Unknown | Unknown |
| Supported embodiments | WidowX, Google Robot, Franka Panda | mobile manipulators, bimanual UR5e, multiple fixed manipulators |
| Training data | 970,000 robot manipulation trajectories from Open X-Embodiment described by the authors. | Robot demonstrations, autonomous data, egocentric human data, and multimodal web data described by the publisher. |
15 comparable fields · 11 material differences · Pair passes the primary-source comparison gate
Primary benchmark evidence
Comparable performance
No protocol-matched benchmark yet.Results appear here only when both models share the same benchmark version, metric, evaluation protocol, and evidence class.
OpenVLA 7B capabilities
Input price—
Output price—
Serving providers0
Canonical IDopenvla/openvla-7b
pi 0.7 capabilities
Input price—
Output price—
Serving providers0
Canonical IDphysical-intelligence/pi-0.7
Internal comparison graph
Related comparisons
Primary evidence