Hey Dan! 👋
Good question. The combo of extensive text amount + a4 dimensions can definitely cause content to get cut off - A4 is a fixed page size, so when the AI generates a lot of text for a card, there's nowhere for the overflow to go.
Here's what's happening: with a4, each card has a strict height limit. When you pair that with extensive (which tells the AI to be really thorough), the generated content can exceed what physically fits on the card.
A few things you can try:
- 1.
Dial back the text amount - Try detailed instead of extensive. You'll still get substantial content, but with less overflow risk on fixed-size formats.
- 2.
Switch to flexible dimensions - fluid or pageless will expand to fit your content instead of cutting it off. These are great for documents where you care more about completeness than strict page formatting.
- 3.
Bump up numCards - If you need to keep A4 and want thorough content, increasing the card count gives the AI more room to spread things out. The content gets distributed across more pages instead of crammed into fewer.
- 4.
Use additionalInstructions - You can add something like "Keep content concise enough to fit on A4 pages without overflow" to give the AI a hint about the constraints.
Quick questions to help me dig in further:
Which API version are you using? (v1.0 or v0.2?)
Can you share your full request body (minus the API key)?
Is the cutoff happening in the generated gamma itself, or only when you export?
If you can grab the x-request-id from the response headers too, that'll help me look into whether this is expected behavior or something buggy on our end.
%20(1).png){kind=avatar}
