Fully local AI media. Everything is made on your Mac, by your Mac. No cloud, no account, no per-use fees.
Every voice, song, image and video is generated on your own machine. Nothing is sent away to be made.
Apple silicon has the memory and graphics to do this. Use what you paid for, without subscriptions or credits.
Nothing you type, record or make leaves your Mac. It works offline once your models are in place.
Each story below began as a few words of text and became a picture, a moving shot, a song and a voice, all on one Mac.
A woman in a dark wool coat stands ankle-deep in the surf on a black-sand beach at blue hour, in the right third of the frame, and every wave that breaks around her feet glows vibrant and bright HDR electric blue. Photoreal cinematic footage. Tiny motes of the same vibrant HDR blue light lift off the water and rise slowly into the air around her like rain falling upward. She slowly raises one open hand and a few motes drift toward it and hover over her palm, lighting her face softly from below. Tall sea stacks stand in the mist behind her, and the left side is calm dark sea and soft mist. The camera dollies in slowly at waist height. Deep blue dusk, cold mist, natural skin texture, shallow depth of field, film grain. Gentle surf and a low, airy hum.
Eight tools that start from words or a recording, and one library that keeps all of it.
Turn writing into natural speech, with built-in voices.
Clone a voice from a short recording, or train one from your own clips.
Transcribe audio and video, with subtitles ready to export.
Write a style and lyrics, and get a full song, sung or instrumental.
Describe a picture and watch it appear.
Change a picture by saying what to change.
Bring a scene to life as a short, moving shot.
Everything you make, in one place. Search it, star it, gather it into projects.
On the way: stem separation, noise reduction, video upscaling, Apple's own voices, and cover art for every song.
One window for everything you make. Each tool opens on what you are about to create.
Reverie Studio checks your Mac's memory and offers the models that fit it. Bigger Macs get bigger models; every Mac gets every kind of media.
Measured on a Mac mini with M4: an image in about three minutes.
Larger image and edit models that follow your words more closely.
Measured on a Mac Studio with M3 Ultra: an image in about 30 seconds.
No account. No tracking. Works entirely offline once your models are in place.
Reverie Studio keeps working in the background, so your smart home, your chats, your scripts and your phone can all make things with it. Everything still happens on your Mac.
Announcements, reminders and Assist replies in any Reverie Studio voice, including one you trained on your own recordings. It listens as well: speech to text over Wyoming, the protocol Home Assistant already speaks. No cloud voice service, no subscription, nothing leaves the house.
Open WebUIMedia from your chatsAsk for an image, a short video or a spoken reply in Open WebUI, and your Mac makes it right there in the conversation.
Make speech, music, images and video from any app, script or shortcut on your network. Send a request, follow its progress, collect the file.
Reach Reverie Studio from your phone over a private network such as Tailscale. Start a song on the train; it is waiting when you get home.
Reverie Studio refines your ideas with the language model you already run. Nothing set up? It downloads a small one that fits your Mac.
llama.cpp
MTPLXAny Mac with Apple silicon, macOS 26 and 16 GB of memory or more.
Get notifiedhello@reveriestudio.app
