A little company.
Mochi is a little mystical animal friend from Prosody Labs. A curious companion who likes music, quiet company, and spending a little time with you.
A talking Mochi friend from Prosody Labs. Speak naturally; you can both listen and talk at the same time.
Your microphone stays on this device. No audio is sent to our servers, recorded or retained. Mochi has no connection to an AI model here.
Conversation mode streams microphone audio to our server. When “Save calls for review” is on, both sides and the model’s generated text are saved privately on Athena for up to seven days, within a 500 MB limit. You can listen back or delete calls in Settings. In chill mode, audio stays on your device. Nothing is recorded or retained, and no audio is sent to the model.
If you add a Spotify player, it connects directly to Spotify. Spotify handles playback and sign-in. For connected playback, authorisation tokens stay in this browser tab; disconnecting clears them. Prosody’s server never receives your Spotify password, tokens or music stream.
Kairos and on-device conversation are in development.
Voice backend loading…
Talk and listen together.
On phones, sensitivity boosts automatically while Spotify plays here. Turn down if Mochi reacts to background noise.
Both sides are saved on Athena for up to 7 days. Listen back or delete them below.
Lower temperatures reduce sampling variation. They are not a quality guarantee. Changes apply to your next call.
Private prototype review. Model text is generated text, not a verified transcript of the audio.
Listen here with Spotify Premium. Your Spotify sign-in stays with Spotify.
Mochi hears music through your microphone. Spotify cannot send its sound directly to Mochi, so music in headphones won’t drive the animation.
Press play in the Spotify player, or open Spotify on your device. Spotify handles playback and any sign-in; some browsers or accounts may get previews. Mochi listens through your microphone, so use speakers for it to hear the music.