Depreciation of the Assistants API by OpenAI, how to replace assistants?

Features
, ,
1

This week, I got the following email from OpenAI telling us that they will shut down the Assistants API within a year. I was just wondering for future workflows that I will set up and also the workflows in which I use the Message an Assistant module how to replace it with another one. It’s important that I can still upload documents.

If I’ve understood their migration document well, you need to replace the assistance for prompts. And I think the place to do so is here: https://platform.openai.com/chat.

After you get a prompt ID, and this prompt ID is accessible via API. Is there already a Make module where you can use this prompt ID, or will there be one in the future?

Earlier this year, we shared our plan to deprecate the Assistants API once the Responses API reached feature parity. With the launch of Conversations, Responses is now at feature parity with Assistants.

We will be deprecating the Assistants API with a planned shutdown date of August 26, 2026 (next year). You will be able to use the Assistants API until that sunset date, but we will not be adding new features or model support to the API.

We recommend migrating your Assistants workloads to the Responses API, which provides a simpler, more powerful foundation for building. Read more in our migration guide.*

Thanks for building with OpenAI. If you have questions about migrating models, please reach out on the OpenAI Developer Forum.*

3

The deprecation caught many of us off guard. I’m transitioning workflows by using Make to orchestrate GPT calls directly via the Chat Completions API. It takes more setup but offers greater control. I also integrate Phonexa into these flows to track lead intent and automate replies—essentially rebuilding assistant-like behavior with more transparency and flexibility.

1 Like
4

@stephenrodgers1990 thanks! This might be a solution: Are you using the HTTP module? Could you share a screenshot of your setup?

I was actually hoping there would be a solution within one of the existing OpenAI modules. I’ve been trying to get it to work with the “Create a model response” module, but for some reason, I never get a good output back, it doesn’t work as smoothly as the assistant module does. Hoping Make maybe comes up with a solution or knows how to properly configure the “Create a Model Response” module. I also couldn’t find much information about it, so it seems like few people are actually using this module. Also see: How to use a prompt ID from OpenAI in your model responses?