R1 Lite used for Physical Intelligence's π0.5
US-based Physical Intelligence built its π0.5 generalist VLA model on the R1 Lite platform.
ImageImage source: opengalaxea.github.io
Open-source dual-system VLA models ('slow thinking, fast execution') released from August 2025; G0 Plus is plug-and-play and G0 Tiny targets edge deployment.
Estimated price
Quote only
Published prices converted at USD 1 = JPY 150, CNY 1 = JPY 21, EUR 1 = JPY 165 for reference. Tax, shipping, duties, support and certification differ by source, so we quote the landed total in Japan.
| Form factor | Robot foundation model |
|---|---|
| Status | Shipping |
Specs are based on maker publications and press reports; real performance depends on configuration and environment. Prices vary with exchange rates, configuration, shipping, duties and support, so we provide final figures by quote.
MAKER
Galaxea AI星海图
🇨🇳 China ・ Beijing ・ Founded 2023
Wheeled dual-arm robots plus open VLA models for embodied AI
Learn moreUS-based Physical Intelligence built its π0.5 generalist VLA model on the R1 Lite platform.
Fei-Fei Li's team released a mobile manipulation kit built on the Galaxea R1 body.
In March 2026 an R1 Pro folded clothes in real time at an eldercare station in Beijing's E-Town.
Validated orders at the thousand-unit level across transport, pick-and-place, packaging, textile folding and equipment linking, with 10,000-unit-level production planned for 2026.
ImageWheeled humanoid with dual 7-DoF arms and an omnidirectional base, covering a wide vertical work range.
ImageGeneral mobile-manipulation platform with dual 6-DoF arms, used as the platform for Physical Intelligence's π0.5.
ImageLightweight force-controlled 6-DoF arm, used with the LEMO desktop dual-arm platform and teleoperation arms for data collection.
Image1X's generalist AI model that lets NEO learn and repeat chores, working with a built-in LLM and memory.
ImageOpen VLA foundation model for humanoids with roughly 3B-parameter weights on Hugging Face.
ImageVLA models; π0.5 targets generalization to unseen homes. Weights and code are released via openpi.
ImagePI's latest model, with better task decomposition, prompt adherence and recovery; a warehouse demo reportedly reached 96.4% autonomy over an eight-hour shift.
Just tell us what you want to do — move parts in a factory, automate inspection rounds. We'll propose robots, a quote and a development plan.
We usually reply within 2 business days