After transforming how AI models are built and shared, Hugging Face is entering the embodied AI world. The company has unveiled two humanoid robots – HopeJR and Reachy Mini, marking its first major step into physical robotics. This follows its recent acquisition of Pollen Robotics, a French startup known for its open-source humanoid platform, Reachy. It was one of the few open platforms in the field, offering modular design, Python APIs, and real-time interaction support.
This acquisition allowed it to build on this foundation. The engineering and design insights from Pollen Robotics now underpin HopeJR and Reachy Mini. The acquisition also helped the company inherit a culture of open innovation that complements its AI values. Additionally, Hugging Face has partnered with The Robot Studio, which co-designed HopeJR, leveraging external robotics expertise while maintaining its open-source philosophy.
With this move, the company is looking beyond chatbots and model hubs. It wants to build real-world systems where AI interacts with the environment, pushing boundaries in robotics while staying true to its open-source DNA.
From open-source models to open-source robots
Hugging Face, ‘GitHub for AI,’ was founded by Clement Delangue, Julien Chaumond, and Thomas Wolf in 2016. Built as a chatbot app targeted at teenagers, it started focusing on creating a platform for creating, testing, and deploying machine learning. It offers a number of data science hosting and development tools, including a GitHub-like hub for AI code repositories, models and datasets. The company offers an open-source library for users to build, train, and deploy AI chat models.
Building on this strong foundation, the company recognised that true AI needs to move beyond software and into the physical world, where AI interacts with real environments. To achieve this, it expanded into embodied AI by robotics. Instead of building closed, commercial machines, it’s releasing robots designed for experimentation and collaboration.
These new robots – HopeJR and Reachy Mini will be available with full documentation, design schematics, and software stacks, just like Hugging Face’s NLP and computer vision models. The complete bill of materials, design schematics, and assembly instructions will be available on GitHub, making these robots truly buildable by the community.
This community-first approach contrasts sharply with the company’s focus on proprietary systems and enterprise customers. Both robots are expected to begin shipping initial units by the end of 2025, and a waitlist for interested developers and educators is already open.
HopeJR: A platform for embodied intelligence
HopeJR, the larger of the two robots, is a full-size humanoid designed for developers and researchers working on AI in physical environments. Equipped with cameras, articulated limbs, and voice interaction capabilities, HopeJR is a modular platform that invites developers to test how language models, perception systems, and control mechanisms work together in the real world. It features 66 actuated degrees of freedom, enabling sophisticated movement including walking and complex arm manipulation. Recent demonstrations show that HopeJR can already walk in simulation, with real-world walking tests as the next milestone.
HopeJR isn’t designed to replace warehouse workers or perform complex field tasks. Instead, it’s a flexible testbed where engineers can explore embodied learning, natural interaction, and task-based robotics. By integrating multimodal models, the company positions HopeJR as a springboard for breakthroughs in human-robot interaction.
Development is progressing rapidly, with real-time prototyping involving design, 3D printing, and assembly being demonstrated just weeks before major robotics conferences. HopeJR is co-designed with The Robot Studio, and its open-source repository is actively maintained, reflecting Hugging Face’s agile and transparent development approach.
Reachy Mini: Compact, accessible, and educational
Reachy Mini, on the other hand, is a scaled-down version of the original Reachy robot from Pollen Robotics. Small enough to sit on a desk, it is aimed at classrooms, hackathons, and indie developers. While its frame is more compact, Reachy Mini includes essential features like expressive displays, basic object manipulation, and integration with Hugging Face models.
The goal is to lower the barrier to entry. Reachy Mini enables hands-on robotics experimentation without the cost, space, or complexity of full-size systems. It’s also an ideal tool for teaching robotics, AI, and human-computer interaction in universities and schools.
Competing visions: Hugging Face vs. Figure AI
The timing of Hugging Face’s robot launch is striking. Just weeks earlier, Figure AI, a Silicon Valley company, announced a $1.5 billion funding round at a massive $39.5 billion valuation. Backed by tech giants like Microsoft and OpenAI, Figure AI plans to mass-produce 100,000 humanoids by 2027. Their robots are designed to work in logistics, manufacturing, and other labour-intensive sectors.
While Figure is pursuing large-scale industrial automation, Hugging Face is targeting researchers, developers, and educators. One builds for factories, the other builds for labs. One is tightly controlled and commercially focused; the other is open, modular, and experimental. These competing strategies highlight a major divide in the humanoid robotics race between proprietary efficiency and open innovation.
While Hugging Face takes an open-source, research-first approach to robotics, it enters a space dominated by high-stakes commercial players, most notably Figure AI. Backed by OpenAI, Microsoft, and Nvidia, Figure recently secured a staggering $1.5 billion at a $39.5 billion valuation, with ambitions to produce 100,000 humanoid robots for logistics, manufacturing, and retail by 2027. Its flagship robot, Figure 01, is engineered for labour replacement, physically capable, AI-integrated, and enterprise-ready.
Figure AI recently ended its collaboration with OpenAI to develop proprietary AI models, citing that OpenAI’s general AI models weren’t suitable for humanoid robotics applications. This move highlights the challenges of applying general-purpose AI to robotics, and underscores Hugging Face’s differentiated open-source approach.
The broader competitive landscape includes Tesla’s Optimus, Boston Dynamics’ Atlas, Agility Robotics’ Digit, and Apptronik’s Apollo, each representing different approaches to humanoid robotics.
This marks a stark contrast to Hugging Face’s newly unveiled HopeJR and Reachy Mini, which aren’t designed for industrial deployment but instead for research, prototyping, and open development. While Figure focuses on closed commercial production, Hugging Face champions community-driven robotics innovation, aiming to democratise humanoid development the way it did with AI models. As Figure races toward scaled automation, Hugging Face is betting on collaboration, modularity, and transparency, a philosophical divide that could define the future of embodied AI.
What lies ahead for Hugging Face robotics?
While the company has not announced pricing or availability dates for HopeJR or Reachy Mini, the company has confirmed that both will be released under permissive open-source licences. Both robots will be supported by benchmark challenges, developer toolkits, and a dedicated robotics section on the Hugging Face Hub, as well as integration with the LeRobot platform, which provides models, datasets, and tools for real-world robotics in PyTorch. LeRobot, launched in 2024, has already demonstrated success with the SO-101 robotic arm, priced at just $100, and serves as the foundation for Hugging Face’s systematic approach to robotics.
By making robotics more accessible and community-driven, the company is laying the groundwork for a new generation of roboticists. Its strategy isn’t about deploying robots at scale, but it’s about accelerating learning, innovation, and collaboration.
CEO Clement Delangue has emphasized that Hugging Face’s open-source approach is intended to prevent robotics from being “monopolized by a few large entities using opaque systems,” reinforcing the company’s mission to democratize AI and robotics.
Final thoughts
The humanoid robotics market is heating up, with billions pouring into commercial ventures that aim to replace human labor. Amid this, Hugging Face is choosing a different path, one that focuses on openness, experimentation, and the belief that the future of robotics should be built in the open.
HopeJR and Reachy Mini may not power factories anytime soon, but they could power the next wave of discoveries in embodied AI. With its deep developer network and open ethos, Hugging Face is uniquely positioned to lead this shift not by selling robots, but by enabling the world to build better ones together.
The educational and research impact of affordable, open-source humanoid robots could be transformative, paving the way for a new generation of innovators to shape the future of embodied AI.
