Reachy and Reachy Mini from Hugging Face
We compare it with our previous Vector and Petoi robots
Hugging Face has had a very busy spring and summer, starting with the release of the SO-ARM101 robotic arm, to the first global hackathon. This week Hugging Face made a big splash by announcing two desktop robots: Reachy Mini and Reachy offered at the price of $299 and $449 respectively. Hugging Face released one main marketing video showcasing the robots potential, which you can see below. Both robots stem from Hugging Face’s acquisition of Pollen Robotics earlier this year. While Pollen Robotics likely didn’t have the firepower to market these robots successfully, Hugging Face being the Github equivalent for the open source machine learning world might have enough ammunition and name recognition to bring this robot into a success.
Specs
The specs of the robot indicate a 6 degree of freedoms head, 2 mobile antennas, full body rotation, 2 microphones, and a wide angle camera. While Reachy Mini is tethered with a cord, Reachy is wireless, has a battery, and has on-board compute in the form of a Rasbperry Pi 5. There are still few details on what could be done on the on-board compute. The marketing videos show that one could write apps at Hugging Face Spaces, and make them work on the robot. Security features, particularly important because the robot has a camera, microphone, and is connected to the Internet are not mentioned.
Our take
While the product is still half-baked (the videos I have seen don’t perform anything more than responsing to music or tracking faces/ hands), I believe that this is finally a robot that would join our existing inventory of Anki Cozmo, Anki Vector, Petoi Nybble. and Petoi Bittle robots.
From a feature pespective, I feel that Reachy is still behind Anki Vector (Although its newer by close to 7 years. Just shows the technical prowess of Anki engineers who made such a great robot). The animations are weak. The security aspects are unclear (Anki worked a ton to make sure that Vector could be securely used at homes, while tethered to the Internet). Reachy is also bigger and heavier, the Reachy Mini is itself a foot in height and weights 1.5 kilogram. The battery life of Reachy is undocumented, and there is no ability to return to a charger when the battery is drained. Which implies that you might be frequently sitting with a charge-less robot on your desk. From this perspective, the robot is certainly half baked compared to Anki Vector which was a mature product even during its first release.
However, there are also a number of aspects which work better for the reachy robots. Just like Vector, Reachy focuses on Human Robot Interactions, and you can certainly see form the videos that the designers want reachy to be emotional. Fast forward 7 years from when Anki Vector was released, machine learning which was at its infancy during Anki days, is now a robust maturing technology. As an example, even the Anki Vector robot is now so much more capable than what it was 7 years ago (Just wishing that Anki would have survived). As an example, take a look at our Anki Vector integrated with the just released Grok-4 from XAI here. There is a lot of interest in robotics thanks to humanoids, than was ever there during Anki days.
An open source robot
Another big difference is that Hugging Face intends Reachy to be a fully open source robot. This VentureBeat article quotes Hugging Face co-founder Clement Delangue mentioning that all aspects of the robot: the firmware, the hardware design, doftware, and applications will be open-sourced. As a start, you could order Reachy un-assembled and assemble it yourself (Hugging Face actualy prefers this and mentions that the first kits they will ship will leverage buyers ability to assemble the robot themselves). This makes Reachy in the league of Petoi robots whose firmware is open source. The aspect of open-sourcing a desktop robot is truly magical, because it implies that lots of robot enthusiasts will openly collaborate to improve this robot. In that sense, the possibilities of this platform are limitless, and you are actually not bound to Hugging Face controlling the robot, or worry if Hugging Face closes shop. The brand recognition of Hugging Face also makes it possible for lots more people to show interest in developing this platform. We can certainly imagine a global hackathon themed on Reachy in 6 months. Vector alas was far from open source, although you can now see lot significant open source activity in Vector’s discord.
Conclusion
Hugging Face’s other co-founder Thomas Wolf indicated in an X-post, that Reachy had already reached $500K in pre-orders. There is also a contest ongoing (refer above X post) where 20 people picked randomly amongst those who have placed orders by Friday will be given a colored Reachy of their choice.
Anki sold about 8000 units of Vector in their Kickstarter campaign which would amount to $1.5 Million. We will see how Reachy fares.
Would you recommend us to buy a Reachy Mini or Reachy? Please use the following poll to vote.
And we also encourage you to help fund our work (and the robots we buy and research) by buying a paid subscription or giving us a one time donation.