Pricing Free & Open
The API is currently free and open. No account, no API key, no credit card. Just send an HTTPS POST to https://api.claudewatermark.org. Fair-use rate limits apply to keep it available for everyone; if you need higher volume or a self-hosted deployment, get in touch.
Open Beta
- No API key required
- All endpoints: text scan, clean, rewrite, image strip
- In-browser playground
- Fair-use rate limits
- Nothing stored server-side
What we do not do
We do not sell a plan that "guarantees" your text will be undetectable. The rewrite endpoint paraphrases text, which reduces statistical watermark and AI-detection signal, but no tool can promise a text will pass every detector, and the statistical token watermark itself can only be read with the model owner’s private key. We would rather be straight with you than invent numbers.
Frequently Asked Questions
Do I need an API key?
No. The API is open. Send a POST request and you get a response. If that changes, this page will say so.
What are the limits?
Text endpoints accept up to 200KB (20KB for rewrite); images up to 12MB. Fair-use rate limiting may apply per IP to keep the service healthy.
Is my data stored?
No. Requests are processed in memory and discarded. Nothing is logged to a database.
Is there a self-hosted option?
The worker is small and self-contained. If you want to run your own copy, reach out.
Why is it free?
The text endpoints are cheap to run at the edge, and the image stripper is pure computation with no upstream cost. The rewrite endpoint uses a small model, which is the only real cost, so it may be rate-limited more tightly. If usage grows enough to matter, a paid tier for high-volume rewrite might appear, but scan, clean, and image strip are designed to stay free.
Can I depend on it in production?
Treat it as a best-effort open service. It runs on Cloudflare's edge and is generally fast and available, but there is no SLA and no support contract behind a free, keyless endpoint. If you need guarantees, self-host the worker, the source is small and the whole thing is a single file per capability.
What about privacy?
Requests are processed in memory and discarded. There is no account, no logging of your text or images to a database, and no third-party analytics on the API itself. That is part of why there is no dashboard: there is deliberately nothing to show.