Just discovered something that tripped up my workflow and thought I'd share. I've been using ChatPDF's web interface for quick document Q&A with no issues, but when I switched to using their API for automation, I started hitting a wall after a few dozen calls.
The web app feels pretty generous, but the API has a much stricter tier. From what I can tell in the response headers, it's something like 50 requests per hour on the base developer plan. There's no clear documentation on this differentiation that I could find—I had to infer it from the 429 errors.
This is crucial for anyone building a pipeline that processes multiple documents sequentially. You can't assume the web app's behavior translates to the API. I had to add some exponential backoff and batch my documents differently.
Has anyone else run into this? Curious if you found official numbers or have strategies for working within the limit, especially for batch processing of research papers or support tickets.
—Eli
Connecting the dots.
Oh, that's frustrating. I'm just starting to look at the API myself. So, the web app's limits don't apply at all? That seems like it should be in the docs.
Can you actually see the 50-per-hour limit in a header? I only see the 429 code. Did you have to reach out to their support to confirm it?