A 299 DOLLARS OPEN-SOURCE DESKTOP ROBOT WITH A FACE THAT RUNS 1.7 MILLION AI MODELS
A 28 cm robot head with six degrees of freedom, expressive eyes, a camera, and direct access to 1.7 million Hugging Face AI models.
RoboticsAI
- difficulty
- ●●●○○
- time
- a few hours
- license
- Apache-2.0
- repo
- repo ACTIVE1,366 stars
●●●○○ · a few hours · Apache-2.0 · 1,366 stars · repo ACTIVE
WHAT YOU’LL NEED
- 3D printer + filament — printable parts — files are in the repo
- Electronic parts — full list with part numbers in the repo BOM
- Dev board / microcontroller — runs the project firmware
- Soldering kit + basic tools — one-time buy, serves every future build
Partner
COMPAREE VERDICT
Reachy Mini is for someone who wants to prototype social robotics or AI agent embodiment without machining a single part. You buy the kit, bolt it together in 2-3 hours, flash an SD card, and you have a desktop head that can track faces, run vision models, and speak. The 1.7 million models claim is real — it is the Hugging Face model hub, accessed through their SDK integration — but most of those models will not run usefully on a Pi 5, and the ones that do still need you to write Python to wire them up. The assembly itself is straightforward: snap-fit shells, screws, and cable routing. The hardest part is not the build, it is deciding what you actually want the robot to do and then writing or adapting the code to make it happen. If you want a platform to learn embodied AI and you are comfortable in Python, this is excellent. If you expect the kit to arrive doing something impressive out of the box, you will be disappointed — the demos exist, but you will spend time configuring dependencies and tuning behavior. The single thing most likely to go wrong is underestimating how much software work stands between you and a demo that feels alive. The hardware is solved; the interaction layer is yours to build.
IN THE REPO
NOT IN THE REPO
- —This is a kit product, not a scratch build — you buy the mechanical parts and electronics from Pollen Robotics, then assemble.
- —The repository contains the Python SDK, simulation files, and documentation. Hardware CAD (STEP files) is linked separately.
- —Firmware for the Orbita actuators and the head controller is in the repo.
- —SDK is Apache-2.0. Hardware files are CC BY-SA 4.0 (permits commercial use with attribution and share-alike).
- —The $299 price is for the wired version. Wireless (with Raspberry Pi 5) starts higher.
- —You will need a computer running Ubuntu 22.04 or 24.04 for the full SDK, or you can run a Docker container on other systems.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
Partner · KickstarterHeyGears G1: 10M+ colors and transparent parts in one print, plus UV printing on flat objects. Figures, parts and labels — no painting.
Videos
Testing Hugging Face's Raspberry Pi-powered open source robot
No official assembly or SDK walkthrough video in the repository. Documentation is text and code examples.
Gallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Order the Reachy Mini kit (Choose wired or wireless version depending on whether you want a tethered setup or onboard Pi 5.)
- 2.Read the assembly guide (Covers mechanical assembly, cable routing, and first boot.)
- 3.Install the SDK (Python package, runs on Ubuntu 22.04/24.04 natively or via Docker on other platforms.)
- 4.Run the Hello World demo (Simple Python scripts to move the head and display expressions.)
Resources
Documentation, files and community threads for this build — we link straight to the original sources and never rehost the creator’s files.
Partner · KickstarterCORE POWER BANK S: 0.2" thin, 2.8 oz, 5000 mAh with 22.5W fast charging. Semi-solid battery, carbon fiber, IPX7 waterproof.
KNOWN ISSUES
- Buying the kit expecting a finished demo robot — you get the hardware platform and SDK, the behavior is yours to code.
- Underestimating the gap between 1.7 million models available and models that run at usable speed on a Raspberry Pi 5.
- Skipping the Ubuntu version requirement — the SDK has dependencies that behave differently on other distros, Docker is the safest cross-platform path.
- Forgetting that the wireless version needs you to supply and configure the Raspberry Pi 5, it does not come pre-flashed.
- Not reading the SDK examples before ordering — if the Python API does not make sense to you, the hardware will sit unused.
- Assuming the Hugging Face integration is plug-and-play — it gives you API access, you still write the Python to load models, process camera frames, and synthesize responses.
What is the difference between the wired and wireless versions?
Wired connects to a host computer via USB, wireless has a Raspberry Pi 5 onboard and runs standalone (still needs WiFi for model downloads and updates). Wireless costs more and requires you to set up the Pi.
Can I run this on Windows or macOS?
Not natively. The SDK is Linux-only (Ubuntu 22.04/24.04). You can run it in Docker on Windows/Mac, but performance and camera passthrough may suffer.
What AI models actually run well on this?
Lightweight vision models (MobileNet-scale), small language models (up to a few billion parameters quantized), and cloud-offloaded APIs. Do not expect to run Llama 70B locally.
Is there a community or forum?
Pollen Robotics has a Discord linked from their main site. The GitHub issues are active for SDK questions.
Can I add arms or a mobile base?
The head is designed to integrate with their larger Reachy robot (which has arms), but Mini itself is head-only. You could mount it on a custom base, the CAD is open.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
1.7 million Hugging Face models are technically reachable, but only a fraction run at interactive speed on a Pi 5. What would you actually build with this — a local agent, a cloud-backed assistant, or something else?
Pollen Robotics
Pollen Robotics is a French company building open-source humanoid robots for research and education. Reachy Mini is the desktop version of their full-size Reachy robot, designed to make expressive robotics accessible without a lab budget.
Keep building — related projects
Same category, picked by how close they are to this build.
FULL CAD3D print this robot arm, show it a task once, and it learns to repeat it
RoboticsAI●●●●○advanced
FULL CADSO-ARM100: The Low-Cost Robot Arm Behind HuggingFace LeRobot
RoboticsAI●●●○○moderate
FULL CADXLeRobot: A Household Robot on an IKEA Cart
RoboticsAI●●●●○advanced
FULL CADYou can 3D-print your own walking Star Wars BD-X droid - fully open-source
RoboticsAI●●●●●expert
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM25d agoedited
The repository is actively maintained, with SDK updates and example scripts added regularly. The hardware files are in a subdirectory, not a separate repo, which keeps everything in one place. The biggest practical gap is the lack of a video walkthrough — the assembly docs are clear, but seeing the cable routing and joint alignment would save confusion. The Hugging Face integration is real but not magic: you get Python bindings to their API, you still write the code to wire a model's output to the robot's expressions or speech. This is a platform for prototyping, not a demo you unbox and show off. If you are comfortable with that, it is one of the most accessible entries into embodied AI hardware.