A screencast showed a forty-character config string I needed. No transcript, no selectable text. I paused, squinted, typed it twice, got it wrong the first time. Same story with locked PDFs, error dialogs that eat clicks, and UI labels that refuse to highlight.
macOS Live Text already does OCR inside Photos, Preview, and Safari. It does not fire on a random region of a Zoom window or a YouTube frame. That gap is the whole product.
Text Lens — one shortcut, Vision framework, done
Text Lens ($6, one-time) is the one I reach for. Sindre Sorhus. Draw a rectangle, text hits the clipboard. Apple’s Vision framework, no account, no network. Eighteen-plus languages, stylized fonts, QR codes. macOS 14+.
If the documents are sensitive, “it never leaves the machine” is the feature. Recognition is under a second for normal UI text.
LensOCR — same job, plus the iPhone as a scanner
LensOCR ($4.99, one-time) is the same draw-to-clipboard loop, then Continuity Camera: point an iPhone at a printed sheet and the text lands on the Mac. No capture history — once you paste, it is gone. Fine if OCR is occasional. The Continuity path is why I would pick it over Text Lens for paper.
TextSniper — languages and read-aloud
TextSniper ($7.99 one Mac, $11.99 unlimited) has been around longer. QR and barcodes, text-to-speech so I can hear a config string while looking at a keyboard, language packs that grow with the OS (Japanese, Korean, Ukrainian, Russian on Ventura+). Multi-Mac licensing is the reason to pay the extra four dollars.
Ghost Text if you will not pay
Ghost Text (Free) is on-device Vision OCR with no global shortcut. Menu bar, draw, paste. For a few grabs a week that is enough. For dozens a day the missing hotkey is the whole complaint.
Live Text first if you only need Preview and Safari. Text Lens if you live in video and dialogs. LensOCR if paper shows up. TextSniper if you work across two Macs or need TTS.