Home
Mingxi Intelligent Open Sources MiniCPM-Robot: 1.5B VLA Model Lets Personal Developers Run Real Robots
Until recently, embodied intelligence was largely the domain of large corporations and research labs—models were too big, entry barriers too high, and most developers never got near a physical robot. But on July 19, OpenBMB broke through that barrier. They released the first series of embodied AI models, MiniCPM-Robot, fully open-sourcing a 1.5B parameter vision-language-action model that lets individual developers run operations and target tracking on real robots. Such openness is rare in the field of embodied intelligence.

This series comprises three core components. The first is MiniCPM-RobotManip, a 1.5B parameter general vision-language-action (VLA) model designed for robot manipulation. The second is MiniCPM-RobotTrack, a compact model for real-world target tracking. Additionally, OpenBMB released PhyAI, a high-performance inference framework built specifically for embodied models. Its goal is straightforward: to let robots truly understand, remember, and act. Together, these three components bring efficient, practical, and open embodied intelligence from research papers and demos into actual robots.
For developers, the biggest advantage is easy accessibility. Model weights and code are hosted in the GitHub repository github.com/OpenBMB/MiniCPM-Robot, and the weights for MiniCPM-RobotManip and MiniCPM-RobotTrack are also available separately for download on Hugging Face. Anyone working with robots can now clone and run them directly.
With the 1.5B VLA model no longer confined to the lab, the pace of embodied intelligence could surpass many expectations.
Project repository: https://github.com/OpenBMB/MiniCPM-Robot
Related article
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber
Related Special Topic Recommendations
Comments (0)
0/500
Until recently, embodied intelligence was largely the domain of large corporations and research labs—models were too big, entry barriers too high, and most developers never got near a physical robot. But on July 19, OpenBMB broke through that barrier. They released the first series of embodied AI models, MiniCPM-Robot, fully open-sourcing a 1.5B parameter vision-language-action model that lets individual developers run operations and target tracking on real robots. Such openness is rare in the field of embodied intelligence.

This series comprises three core components. The first is MiniCPM-RobotManip, a 1.5B parameter general vision-language-action (VLA) model designed for robot manipulation. The second is MiniCPM-RobotTrack, a compact model for real-world target tracking. Additionally, OpenBMB released PhyAI, a high-performance inference framework built specifically for embodied models. Its goal is straightforward: to let robots truly understand, remember, and act. Together, these three components bring efficient, practical, and open embodied intelligence from research papers and demos into actual robots.
For developers, the biggest advantage is easy accessibility. Model weights and code are hosted in the GitHub repository github.com/OpenBMB/MiniCPM-Robot, and the weights for MiniCPM-RobotManip and MiniCPM-RobotTrack are also available separately for download on Hugging Face. Anyone working with robots can now clone and run them directly.
With the 1.5B VLA model no longer confined to the lab, the pace of embodied intelligence could surpass many expectations.
Project repository: https://github.com/OpenBMB/MiniCPM-Robot
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber











