Possible Bug? Open AI Assistant Does Not Work, Chat Completions Does

Hi all,

I’ve been loving the OpenAI perk and have found that it works great with the chat completions endpoint, but it does not work with the assistant endpoint.

The issue happens whenever I attach a file (e.g., .pdf or .docx). The assistant doesn’t generate any results or does not recognize there is a file attached. I tested the exact same setup using OpenAI’s Playground, and it worked perfectly with the same assistant and user prompt I’m using in my Gravity Form. This leads me to believe there might be a bug with the assistant endpoint when attaching files.

When I upload an image to it, I do receive a message stating that it only accepts files—which makes sense, as this is what OpenAI chat completions should be used for. However, when I use the Gravity Wiz OpenAI Assistant endpoint with a file, it doesn’t work at all.

I’ve attached my full setup for reference.

Note - My assistant endpoint is using OpenAI’s latest API if that has anything to do with it. It will not let me role back to v1 API to test.

Is this a known issue? Can anyone else confirm if .doc or .pdf files are generating results for them?

Thanks everyone in advance!

I believe there may be a bug.

Full setup:

Assistant endpoint I am using:

This Shows it works in my playground:

Thanks for your help! It only allows me to upload one image at a time so excuse all the replies. :slightly_smiling_face:

It does submit the file but does not provide a response:

Hi John,

Based on those settings, I would expect the summary to be mapped to the Answer field once the form is submitted. I tested an Assistant with the same setup, and it returned a summary.

I think this is going to require further investigation from us. Could you open a support ticket so we can take a look?