FeaturesRead text
Copy the text out of any screenshot
Text you can see but cannot select: a chat message, an error dialog, a scanned letter, a slide on a call. Read text transcribes it exactly as it appears.
Runs on the Swift tier — 1 credit per run.
What you get back
The text and nothing else. Reading order, line breaks and paragraphs are preserved; capitalisation, punctuation, spelling and accents are kept exactly, typos included. Tables come back as tables, lists as lists, code in a fenced block with its language. A character it genuinely cannot make out is marked [?] rather than guessed. It does not translate, correct or shorten anything.
When to reach for it
Anything that will not let you select it: a video frame, a photo of a whiteboard, a PDF that is really an image, an app that blocks copying, a message you want to quote precisely. It runs on the lightest tier, so it is the cheapest lens there is.
What leaves your machine
Exactly one image, once, when you press the lens — plus the capture's window title and dimensions, nothing else. It is read by a private model under zero data retention, the answer streams back as text and is saved onto the capture in your local library, and the image is not kept anywhere.
How to use it
- Capture — PrtScn for a region, Alt+PrtScn for a window, Ctrl+PrtScn for the whole screen.
- Pick the lens in Studio. The answer streams in and stays attached to that capture.
- Copy it, rerun it, ask a follow-up, or come back to it later from the Library.
Every lens on every plan
The free plan runs every lens, with a monthly credit allowance and nothing to configure. Capture, editing, redaction, the library and the presenter tools need no account at all.