Reachy‑Motion

How it works

1. 1You talkJust speak to the robot — it listens through its own microphone.
2. 2It answers out loudA live voice model replies, fast enough for real back-and-forth.
3. 3Every clause becomes a gestureAn AI choreographer turns each phrase into head, antenna and body motion, timed to land with the audio.

On top of that it breathes, sways gently while it speaks, and reacts the moment you start talking — no frozen robot while the AI thinks.

New in v0.4: switch the motion engine to Binh Pham's flow-matching motion model for learned, organic movement — or run his fine-tuned planner and the model entirely on your Mac, no cloud needed.

Pick a voice

Switch any time from the app page — or just ask the robot: “switch to ElevenLabs”.

Things to try saying

• “Look at me!”
• “What do you see?”
• “Turn it down a little.”
• “What's the weather today?”
• “Any news this morning?”
• “Stop staring at me 😅”
• “Tell me a story — act it out!”
• “Switch to Taiwanese.”

A small “body agent” understands these by meaning, not keywords: it moves the camera to follow your face, takes a picture to describe what's in front of it, adjusts the volume and looks things up on the web.

Install in three steps

1. Find it in your dashboard. Open Reachy Mini Control (or the robot's dashboard), go to Discover apps, search for reachy_motion and click Install.
2. Add your keys. Start the app, open its settings page and fill in Keys & voices: an OpenAI API key, and for ElevenLabs mode your agent ID.
3. Say hi. Pick a voice and start talking. Reachy will take it from there.

Works on Reachy Mini Wireless (runs on the robot itself), Reachy Mini Lite, and the simulator.