Point it at a block of text — on your screen, in a PDF, in a photo of a whiteboard — and it comes out the other side as real, pasteable text in about a second, read right off the page with Apple's own on-device OCR.
macOS 13+ · Screen Recording permission only — no Accessibility, no Input Monitoring
The capture loop, history, Capture Viewer, Markdown mode, read-aloud, and QR/barcode detection all run entirely on-device with Apple's Vision framework. No account, no sign-in, no network call — unless you explicitly turn on Handwriting Expert.
A single customizable hotkey works from inside any app, so you never have to switch to Unbox Text first. Change it any time in Settings — it checks for collisions with the app's own other shortcuts as you record a new one.
A quick walkthrough of the capture loop, history, and a few of the modes below.
Press your hotkey (⌘⇧2 by default) from inside any app, drag a box around whatever you want — a sentence, a paragraph, a whole screen — and Unbox Text reads it with Apple's on-device Vision OCR and drops it straight onto your clipboard, usually in under a second.
Nothing leaves your Mac to make that happen, and there's no account to sign into first. Don't like the default shortcut? Change it any time in Settings.
Drag a PDF or image straight onto Unbox Text, or use File → Import, and it runs the same on-device recognition over the whole file instead of just what's visible on screen.
Multi-page PDFs get a page-range picker, so you can pull text from page 12 of a 200-page document without waiting on the other 199.

Unbox Text keeps a history of everything it's ever recognized, each one with a thumbnail of the original image or PDF page right next to the text it found.
Search across the whole history, reopen any past capture, and re-copy or re-export it whenever you need it again — nothing you've captured is ever really gone.

Open any capture in the Capture Viewer and you get three panes side by side: the original image, the text Unbox Text recognized, and that same text with low-confidence words highlighted.
Click a highlighted word to correct it right there, before you copy or export — so a single misread character doesn't ride along into whatever you're pasting into.

Turn on Markdown mode and Unbox Text does the formatting work for you: it spots code blocks by the geometry of the text itself — consistent monospace character widths, not keyword guessing — and wraps them in fenced code blocks automatically.
Every capture can be read aloud. A small "now playing" pellet floats on screen across apps and Spaces, with play/pause, skip forward or back by sentence, a scrubber, and a voice and speed you pick.
Turn on auto-speak and Unbox Text starts reading the moment a new capture finishes — handy for a long PDF you'd rather listen to than scroll through.
Catch a QR code or barcode in a capture and Unbox Text decodes it on the spot — no separate scanner app needed for something that was already sitting on your screen.
Recognition isn't limited to English either: it uses Vision's own language auto-detection, so text in most of the languages Vision supports gets read correctly without you telling it which one to expect.
On-device Vision is good, but cursive or messy handwriting is still genuinely hard for any on-device model. Handwriting Expert is an opt-in mode, off by default, that routes that one capture through Microsoft Azure's cloud Read API instead — which does meaningfully better on handwriting.
It's bring-your-own-key: you create your own free Azure account (the free tier covers around 5,000 reads a month) and paste your own endpoint and key into Settings. Unbox Text never sees, stores, or pays for that usage — it's your account, your quota. A one-time disclosure names Azure explicitly before this can ever be turned on.
Everything else on this page stays on-device, with or without Handwriting Expert turned on.
Check the support page first — if it doesn't cover it, we'd like to hear about it directly.