Identity
- Status
- active
- Developer ID
pi-0.7- Released
- Apr 16, 2026
- Version
- 0.7
- License
- Unknown
Physical Intelligence's steerable generalist robotics model for dexterous, cross-embodiment task execution.
0 current hosted offerings
No exact current provider endpoint or published price has been verified for this model version.
At a Glance
pi-0.7Publisher Media
Embodied Intelligence
Robot demonstrations, autonomous data, egocentric human data, and multimodal web data described by the publisher.
Verified Platform Links
This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Primary Evidence
Comparable Peers
| A | Pair | B | Context |
|---|---|---|---|
pi 0.7Physical Intelligence | vs | MolmoAct 2Ai2 | cross-developer peersrobotics |
pi 0.7Physical Intelligence | vs | Helix 02Figure | cross-developer peersrobotics |
pi 0.7Physical Intelligence | vs | Gemini Robotics 2Google DeepMind | cross-developer peersrobotics |
pi 0.7Physical Intelligence | vs | SmolVLA 450MHugging Face LeRobot | cross-developer peersrobotics |
pi 0.7Physical Intelligence | vs | GR00T N1.7 3BNVIDIA | cross-developer peersrobotics |
pi 0.7Physical Intelligence | vs | OpenVLA 7BOpenVLA Research Team | cross-developer peersrobotics |
Questions
Physical Intelligence's steerable generalist robotics model for dexterous, cross-embodiment task execution. It is developed by Physical Intelligence and its current sourced lifecycle status is active.
pi 0.7's current record lists Text, Image, and Robot state as input and Robot action as output.
pi 0.7 is classified as a vision-language-action model. Its action representation is Continuous robot actions conditioned by multimodal prompts, its control architecture is High-level policy, world model, and action expert, and inference is recorded as unknown.
The developer reports mobile manipulators, bimanual UR5e, and multiple fixed manipulators as supported or represented embodiments. No direct model-to-platform relationship is currently verified in the catalog.
The current record lists an unknown total parameter count and an unknown active parameter count. Its architecture is High-level policy, world model, and action expert.
pi 0.7's recorded capabilities are cross-embodiment, dexterous-manipulation, language-steering, and visual-subgoals. Its supported tasks are robot-control, dexterous-manipulation, and mobile-manipulation.
Robot demonstrations, autonomous data, egocentric human data, and multimodal web data described by the publisher.
pi 0.7's current primary-source record does not establish API availability. No provider endpoint is treated as available without direct provider evidence.
No directly sourced token price is currently attached to pi 0.7. Missing prices remain unknown rather than being inferred from another model or provider.
pi 0.7 is not marked as an open-weight model. Self-hosting is not established by the current source, with no license recorded.