AI
Apple Intelligence for your pictures and PDFs
Ask a picture a question, find pictures by what is in them, lift a subject off its background, or get the key points of a long PDF with the page each comes from. It all runs on your Mac, with Apple's own models and Vision. No key, no credits, no server.
- Runs on your Mac
- No AI key needed
- Nothing uploaded
- Apple Intelligence and Vision
What it does
Tools you reach for, not a layer over everything
Each one starts when you ask for it, and works on the picture or PDF in front of you.
Ask about a picture
What is in it, what a sign says, what a chart shows. Answers stream in as they are written.
⌘K, or Ask in the viewer's toolbar.
Ask a PDF, or summarise it
The key points of a long document, or an answer to your question, each with the page it comes from. When the document does not say, Pixo says so.
Ask in the PDF toolbar, then Summarise or a question.
Search by what is in the pictures
Type "receipts", "passport" or "dog on the beach". Each picture is read once on your Mac, then searches are instant.
The search field, set to Contents.
Lift a subject
The subject off its background. Click to choose which subject when there are several; copy, save or drag it out.
Lift Subject in the viewer, or ⇧⌘L.
Best shots and duplicates
Groups pictures that look alike and picks the sharpest, best-composed one of each. The rest can go to the Trash in one step.
Image › Find Best Shots, or Find Duplicates.
Rename and describe
File names from what is in a picture or PDF, and a description saved into the picture for alt text and Spotlight.
Image › Rename with AI, and Describe in the inspector.
Crop to subject
Square, 4:5, 16:9, 9:16 story or 3:2, placed around what draws the eye.
Tools › Crop to Subject.
Invoices to a table
Date, vendor, number, total and due date from many invoices and receipts into one CSV. A value is kept only if the document really contains it.
Tools › Invoices to Table.
Translate, tables, Reimagine
Translate the text in a picture with Apple Translation, copy a table in a picture as CSV, or open it in Image Playground.
The Image menu.
How it works
Your picture goes to your Mac's own model. That is the whole trip.
Apple Intelligence runs on Apple silicon, inside macOS. Pixo asks it, and Vision, the way any Mac app can, and nothing is uploaded on the way.
The picture or PDF
You ask a question. Pixo reads the picture with Vision, or finds the pages of the PDF that answer it.
Apple's model, on your Mac
Pixo gives that to Apple Intelligence on your Mac. On macOS 27 the model also sees the picture itself.
The answer, in Pixo
Shown next to the picture, with the page numbers for a PDF. Nothing was uploaded.
Not on the path: pixoviewer.com, or any cloud. Your pictures, your PDFs and the questions you ask stay on the Mac.
What you need
- 1
macOS 26 on Apple Silicon
Apple Intelligence runs on Macs with Apple silicon. On an Intel Mac Pixo works, and says plainly that these features need Apple silicon.
- 2
Apple Intelligence turned on
In System Settings › Apple Intelligence & Siri. Pixo says so plainly if it is off or still getting ready.
- 3
Nothing else
No account, no key and no download inside Pixo. Vision tools such as Lift Subject work even with Apple Intelligence off.
You stay in charge
How Pixo keeps the answers honest
The model on your Mac is small and quick. These rules are built into the app so you can check what it says.
Answers name their page
An answer about a PDF says which page it comes from, so you can look before you rely on it.
It says when the document does not say
Asked about something a PDF does not contain, Pixo answers that it is not there, instead of guessing.
Figures are copied, not worked out
Amounts, dates and numbers are taken as written. In a table from invoices, a value the document does not contain is left empty.
Your originals are never changed
Renaming asks first, describing writes only the description field, and every other tool saves a new file.
What it costs
Part of Pixo, with nothing to pay per use
Free
10 questions a day about a picture, Lift Subject, translating the text in a picture, copying a table as CSV, Reimagine, and finding best shots and duplicates.
With Pro: $19.90 per year or $29 once
Questions without a limit, search by content, renaming and describing with AI, crop to subject, moving duplicates to the Trash, PDF summaries and questions, and invoices to a table. Apple runs the model, so we sell no AI credits.
See pricingLimits
What this page does not promise
The model is small. Apple's on-device model is quick and private, not a large cloud model. It can miss a point in a long document or misread a blurry picture.
Long PDFs are read in parts. For a long document Pixo picks the pages that best match your question. A point spread over many pages may need a narrower question.
Languages. Apple Intelligence answers in the languages Apple supports on your Mac.
Not every Mac. Apple Intelligence needs Apple Silicon. On an Intel Mac with macOS 26, Ask, PDF summaries, Rename with AI, Describe, Invoices to Table and Reimagine are not available; Live Text, Lift Subject, Best Shots, Duplicates, Translate and search by content still work, the search with simpler keywords.
Questions
About AI in Pixo
Where does the AI run?
On your Mac, with Apple's own models (Apple Intelligence) and Vision. Pixo sends your pictures and PDFs to no server, ours or anyone else's, and sells no AI credits.
Can I trust the answers?
Mostly, and check what matters. Answers about PDFs name the page they come from, figures are copied as written, and when a document does not say something Pixo says so. The on-device model is small: for a contract or an invoice, look at the page before you rely on a number.
Does Apple Intelligence see my picture?
On macOS 27, yes: the picture itself is given to Apple's model on your Mac. On macOS 26 the model cannot take pictures, so Pixo first reads the picture with Vision (what is in it, and its text) and gives the model that. Answers about colours and composition are better on macOS 27.
Can Pixo tell if a picture was made by AI?
It shows what the file says about itself: Content Credentials (from ChatGPT, Adobe Firefly and others), the IPTC "made by AI" tag, and prompts saved by Stable Diffusion, ComfyUI, Midjourney and others. Many AI pictures carry no such tags, and tags can be removed, so a picture without them may still be made with AI.
Ask your own pictures.
Press ⌘K.
Download Pixo and press ⌘K on any picture. It needs macOS 26 or later on Apple Silicon, with Apple Intelligence turned on.