Khint vs Cleanshot X,OCR with AI on top.
Cleanshot annotates screenshots. Khint pulls out the text and runs AI on it.
Eleven capabilities,
two captures.
OCR engine, annotation, recording, AI follow-up, pricing. The honest side-by-side.
What you can do with the captured text.
Khint: AI actions
Capture a stack trace, an error from a screenshot, a UI label in a foreign language, a blob of text from a PDF. Hit Cmd+Shift+K and run "Explain this error", "Translate to English", or "Summarize for a ticket". The text never leaves the device until you trigger the Action.
Cleanshot X: annotation
Add arrows, callouts, redact sensitive regions, blur faces, build scrolling captures, record GIFs, share via Cleanshot Cloud links. The right tool when the deliverable is the screenshot itself, not what's written on it.
Pricing: one-time $30 vs free + €9.99/mo.
Cleanshot X is a one-time $30 license (also bundled in Setapp). Khint is free for 10 AI actions per day and 5 OCR captures per day, and €9.99/month for unlimited AI actions and 50 OCR captures if you exceed the free tier. If you want only OCR occasionally, Khint's free tier covers it. If you want only annotation, Cleanshot is the cleaner buy. Together, the bill is $30 + €0/month for casual AI users.
Can I use both?
Yes, and many users do. Cleanshot on Cmd+Shift+5 for annotated screenshots and recordings. Khint via the palette (Cmd+Shift+K) when you need the text rather than the picture. They don't overlap on the actual jobs-to-be-done.
Common questions.
Does Khint replace Cleanshot X?
Not for everyone. Cleanshot X is the reference for Mac screenshot annotation: arrows, blur, callouts, scrolling capture, screen recording, share links. Khint does none of that. What Khint replaces is the Cleanshot OCR tier: the palette shortcut captures screen text and you can immediately run AI actions on the captured text. If your job is annotated screenshots and recordings, keep Cleanshot. If your job is pulling text out of a screen and doing something with it, Khint is enough.
Can Khint annotate screenshots?
Not today. Khint's OCR pillar is text-only: it captures pixels, extracts text, and lets you act on it (paste, send to an AI Action, save to a session). Drawing arrows, callouts, blurring sensitive regions, or producing share-ready PNGs is not in scope and not on the short-term roadmap. Cleanshot remains the right tool for annotation.
Does Khint OCR run offline?
Partially. On macOS, Khint captures the screen on-device — the screenshot stays on your Mac. The image is then sent to Anthropic's Claude Vision API to extract the text; nothing is stored beyond that single call. You only need an internet connection if you then trigger an AI Action on the captured text, which calls the Claude API.
Which has better OCR accuracy?
Both run accurate OCR on standard screen text. Cleanshot's OCR ships in a paid bundle. Khint's OCR ships in the free tier (5 captures/day) and integrates the text directly into the same palette you use for AI actions, so there's one less paste step.
Is there a free Cleanshot alternative with AI?
Yes. Khint is a free cleanshot ocr alternative that focuses on screenshot to text mac and AI follow-up rather than annotation. The free tier covers 10 AI actions per day and 5 OCR captures per day, with unlimited local history. If you outgrow the free tier, Premium is €9.99/month for unlimited AI actions and 50 OCR captures per day.
Capture screen text. Act on it.
Screen-to-text OCR via a keyboard shortcut. Free, with 10 AI actions per day on top.