Video
Raw clips in. A graded film out.
Auto Cinematic finds the best moments, cuts on the beat of your scenes, grades the color and letterboxes it in 2.39:1.
2 min 20 s of 1080p → a 30 s film in 17.1 s
Aura Compute presents Atlas
Make anything. Keep everything.
Atlas is an AI creative studio that lives on your own computer. Film, photos, images and words, made at the speed of your ideas, and never out of your hands.
- 17 s
- to cut and grade 2 min 20 s of footage
- 0.19 s
- to perfect an 8 MP photo
- 0
- tokens counted, on any plan
Preview Atlas by Aura
One studio for everything you make.
Film, photos, images, words and ideas. Atlas does the heavy lifting on your own chip, so your files stay with you and your ideas never wait in a queue. KIRA keeps you company. GlassBox checks the facts.
Preview on Apple silicon Macs. Windows and Linux in development.
Why does light bend near a black hole? Show me.
Mass curves spacetime, and light follows the straightest path through that curve [1]. Near the event horizon the bending becomes extreme; at 1.5 times the Schwarzschild radius, light can orbit [2].
GlassBox · 2 claims · 2 supported
[1] NASA Science[2] Wikipedia
Photo
Fixed, and it shows its work.
Auto Enhance corrects light, color and crop, then tells you exactly what it changed, with Undo.
8 MP photo in 0.19 s
Images
Made on your machine.
Describe it and Atlas draws it locally. Your prompts are not sent anywhere to be rendered.
1024 × 1024 in about 37 s
Words & ideas
Write, research, understand.
Draft anything, get live answers with numbered sources, and turn explanations into visuals you can touch.
First word in 0.35 s
Intelligence, closer to you
The most powerful computer in your life may soon know what you remember, what you create and how you think. If intelligence becomes that personal, it should stay yours.
Speed
Fast, because there is no trip.
Atlas starts answering about as fast as the quickest cloud models, without your words leaving the room. With photos and footage it pulls far ahead, because nothing has to upload. Our numbers, measured on one machine and published with the method.
0.35 sto the first wordqwen3.5:4b, warm, median of 10
17 sto cut and grade 2 min 20 s of 1080pAuto Cinematic, about 8× real time
0.19 sto auto-enhance an 8 MP photofull-resolution JPEG, median of 5
0 sspent uploadingyour files are already on your disk
Measured September 29, 2026 on one Apple M5 Pro MacBook (15-core CPU, 16-core GPU, 24 GB), on AC power, Ollama 0.34.3, everyday apps left open. See every benchmark and how to reproduce it →
Rent or own
Cloud AI is a taxi meter. Aura is the car in your driveway.
Every question you send to a cloud AI runs on someone else's GPU, in someone else's building, on someone else's bill. That is why it is metered, rate-limited and far away. Aura runs on the chip you already paid for.
- Your request leaves your computer
- It waits its turn on shared GPUs
- Every token is billed, so plans have limits
- Your request stays on your computer
- The model is already warm in your memory
- Nothing to meter, so no token limits
Close to you
It knows your work because it lives where your work lives. Your files, notes and footage stay on your machine.
No telemetry. No automatic cloud fallback.
Close to your wallet
No token limits on any plan. Your computer does the thinking, so there is nothing for us to meter.
Close to your time
Nothing to upload and no queue to join. A clip starts editing the moment you drop it in.
Close to the planet
The chip in your laptop is already on. Everyday AI should not need a round trip to a data center.
We will publish energy numbers once we have measured them.
The personal intelligence stack
Not a collection of apps. The layers personal intelligence needs.
Follow one request down through the stack and back up to you. Every layer runs on your machine.
- 01KIRAIn Atlas
You ask.
Out loud or typed. KIRA hears you and shows that Atlas is listening. Intelligence should not feel like another dashboard.
- 02ATLASPreview
Atlas works out what you need.
An answer, a live search, a visual, an edit. It plans the work before any model runs.
- 03TRACEIn development
Trace remembers.
What you saved, and why it mattered. Your own context, brought back with your permission.
- 04OMNISCOPEResearch
OmniScope looks.
If there is an image or footage, it sees it. It can tell a letterbox bar from a genuinely dark scene.
- 05NUCLEUSResearch
The model reasons.
Today Atlas runs open-weight Qwen 3.5 models. Nucleus is our research into smaller, sharper ones. Bigger is not the only path to better.
- 06BARX 2.0Shipping in Atlas
Barx runs it on your chip.
It reads your machine and picks the model size, memory, threads and GPU layers that fit right now. The decision takes under a millisecond.
- 07GLASSBOXLite in Atlas
GlassBox checks, then you get the answer.
Each claim is checked against its sources and marked supported, partial or unsupported. You see the verdicts, not just a confident tone.
Hi, I'm KIRA. I live inside Atlas.
In Atlas KIRA
Less dashboard. More presence.
KIRA is how Atlas feels. A companion with thirteen gestures and real states: listening, thinking, speaking, creating and resting. Serious technology can still be warm.
Pricing
One flat price. No token limits.
Cloud AI charges for every word because every word costs them a GPU. Atlas thinks on your hardware, so we can charge like software, not like a taxi.
Atlas
For everyday creating
$20/ month
- Unlimited chat, live research and visuals
- No token limits
- Studio, 2 sessions a day: Auto Cinematic video, photo Auto Enhance, image generation
- KIRA, GlassBox checks on every answer, Barx runtime
Atlas Studio
For creators who make every day
$100/ month
- Everything in Atlas
- No token limits
- Unlimited Studio: auto video editor, photo editor, image generation
- Agent access Coming soon
- Auto co-work Coming soon
Planned launch pricing in US dollars. We are not taking payments yet.