API Discussion

Does the API utilize the Gamma-5 engine now and if not, is there any timeline on updates there?

Hey team, posting this in API discussion because it's a developer workflow issue. When using POST /v1.0/generations, the first card often returns image-layout="behind" and an empty accent image slot, regardless of what's in the request. Every subsequent card correctly returns image-layout="blank". The empty slot renders as a gray band above the content on the first card, which is visible to end users of documents we generate programmatically. Our setup: format: document, textMode: preserve, cardSplit: inputTextBreaks, cardOptions.dimensions: pageless, imageOptions.source: noImages. We are not using GML. We are on the plain text and additionalInstructions path. We have tried explicit additionalInstructions wording telling the generator not to use a cover layout on the first card. Results are inconsistent. What we are asking for is a request parameter such as firstCardLayout: "blank" or suppressCoverLayout: true that tells the generator to treat the first card identically to all other cards, with no accent image slot. Alternatively, honoring imageOptions.source: noImages consistently on the first card the same way it does on all subsequent cards would solve it. This comes up across virtually every document we generate and currently requires a post-generation patch call to fix. A single parameter would eliminate that entirely.

Hi — two API limitations that are blocking us, both around adjusting presentations after they've been generated. Context: we generate presentations programmatically at volume through the API, as deliverables inside our own product. PROBLEM 1 — no way to edit an existing presentation via the API The API creates presentations but can't update them. So when we improve our house style, or spot a layout problem in something already generated, we have two options: regenerate it, which produces a new ID and URL and breaks the link already embedded in our product, or open each one in the editor and fix it by hand. For a handful that's fine. At our volume it isn't, and it means any styling improvement we make can only ever apply to future output, never to what we've already shipped to clients. What would solve it: an endpoint that applies an instruction to an existing presentation in place, keeping the same ID and URL — essentially the in-editor AI edit, exposed via the API. Even a restricted version limited to layout and styling, with content explicitly untouched, would cover most of our need. We've tested the equivalent prompt manually in the editor and it does exactly what we want; we just can't reach it programmatically. PROBLEM 2 — fixed card sizes shrink content instead of giving it more room, and we can't tune that from the API We need 16x9, because our clients export to PowerPoint. But with a fixed card height, dense slides get scaled down to fit rather than spread out, so text ends up small and cramped. Our worst example is a 25-slide presentation with five tables where almost every slide is visibly squashed. The auto-fit behaviour doesn't appear to be exposed as an API parameter at all — cardOptions only offers dimensions and headerFooter — so we can't influence it at generation time. Related: textOptions.amount is documented as applying only when textMode is generate or condense. We use preserve, because our copy is client-approved and must not be reworded. That leaves us no way to express "same words, spread across more slides." What would solve it: either expose the auto-fit / content-scaling behaviour as a generation parameter, or make textOptions.amount (or an equivalent density control) take effect alongside textMode: preserve. We know numCards and cardSplit: inputTextBreaks still work under preserve and we'll use them — but both require us to guess the right slide count up front, rather than letting Gamma lay content out well for the card size we've chosen. Happy to share specific examples if useful.

Hey Arihant J., is this still happening for you? Do you have a simplified steps to reproduce that reliably reproduce the issue?

Hello folks, I'm using Gamma APIs to generate PPTs from text, and I'm seeing major issues around AI image generation, like garbled text and malformed images. ❌

I did find some success with using stock images and illustrations ✅ Wondering if anybody else here has faced similar issues, and if there is a workaround for this?🤔

Hi Gamma team, I am running into an issue using templates with the API. In short, I've created a template, and I need to ensure that I am sending the verbatim text forward for the presentation that I'm creating. All of the research points to the fact that "preserve text" is not an option that can be used with the template's endpoint. Is there any workaround for this? The actual verbatim text that is being pushed forward is absolutely critical. So, as a little bit more context, I ended up going down the templates path due to the 5,000-character limit associated with additional instructions, which I was using to define and describe each of the slides.

Hi 👋 I have been using Zapier to receive email requests to generate gammas. Roadblocks I experience are: (1) when the sender is another person other than the Gamma owner, they get a shared access email but then cannot access it even if they try to sign in is what I hear--could be user error. (2) The image below shows it the external access as view with the email recipient so maybe sign in is not required. (3) If I remove the email recipient access in the field to test, it publishes but then I get an error essentially saying it is required. (4) If I as the owner of the gamma do the zap it does not send me an email saying it completed (or shared with me or anyone--only when someone clicks it).

(5) I also have the PDF export as toggled on and not sure that is retrieved or doing anything. Thanks for your help! 🙏

Feature Request: Clickable Hyperlink Support in inputText (preserve mode) We are using the Gamma API to generate professional research documents for investors. Our workflow assembles a large body of market research content - sourced from live web research - and sends it to Gamma via the generations endpoint using textMode: "preserve". At the end of every document we include a Sources section that lists the citations used throughout the document. Each citation has a label (e.g. [1.1], [4.3]) and a source URL. We want readers to be able to click those URLs to open the source in a browser. What we tried: - Standard markdown link syntax: [1.1](https://source-url.com) - Gamma renders this as literal text, not a clickable link. - Bare URL: https://source-url.com - not auto-linked. - After reviewing the full developer documentation at developers.gamma.app, we found no documented syntax for creating clickable hyperlinks within inputText. What we are asking for: A supported way to include clickable hyperlinks in inputText content when using textMode: "preserve". This could be standard markdown link syntax [text](url), bare URL auto-linking, or a Gamma-specific encoding - whatever fits the platform's architecture. A secondary related request: per-card textMode control, so that specific cards (such as a References card) could use textMode: "generate" while the rest of the document uses textMode: "preserve". This would allow Gamma's own rendering to handle links on those cards. Why it matters: Source citations are a credibility requirement in professional research documents. Without clickable links, readers must manually copy and paste each URL - which is a significant friction point in a document designed to build investor confidence.

Question / comment regarding the API: As I attempt to list themes, there appears to be a pagination issue both in the Gamma node (which is a community node in n8n) as well as when using an HTTP node to access the API. It will only list the themes from the first page, and if I have any custom themes beyond page one, I cannot access them. I did find a workaround with the help of Claude code, but truncation issues can be challenging, especially if you do not recognize it...

Hi All & Max J. I'm currently on 75£/month for 20,000 credits using only API is it possible to increase the subscription to get a better economies of scale on price per credit? I think I would need probably 40,000 month. The current problem is that once you go out of credits is incredibly expensive to top up, and I believe Im already on the highest paid subscription?

hi is there any way to completely opt out of the ai infographics - they look really awful in most of our presentations....

When you want to generate images in different slides and you are using inputText linbreak --- what is better approach?

(eg. --- slide 1 text etc... Image: description of teh image that you want to build ----)

(e.g Additional Instructions: -Always add related image background on slide 2 and 4) I've seen sometimes it behaves inconsistently so im wondering if any has come up with a solution that is the most deterministic to tell Gama how to behave in terms of images for specific slides

Being able to edit via api would of course be great, but I can already change permissions for all api generated content to have permissions to edit/comment, so if I changed that permission, my users get the link we provide via api, and then edit those gammas. The only negative is then after they are done, and then they want to share the link with their clients, those clients have the same permission to edit. It would be great if once they edited, they could have the ability to change the permissions to view only. It would make sense if that was something I could allow them to do from the main account page that host the api's.....'allow link recipient to change permissions (from edit/comment to view only)' It would also drive a lot of people into the gamma experience to edit since they have to set up at least a free acct....so there's a benefit, too

+1, generated gamma name parameter instead of first characters of prompt would be incredibly helpful !

Hey, we have PDFs being created via Gamma, and then stored to a bank of knowledge. These files need to be referred to easily by agents on our platform, and file names would have a large weighting to allow this. Gamma currently creates these file names by using the title only (we believe) of the presentations it creates. However, sometimes the title is very short/non-descriptive and the subtitle/other headers have the vital context. Therefore, I was wondering if there is a better way to control the file names of the assets created by Gamma. Maybe a parameter we can include while making the generation call to Gamma for the file_name of the asset? That way we can get our ai to come up with the best name for it. Or is there any way that the file_name created by Gamma can use that extra context from the subtitle if the title is short