OpenAI Agent Builder, Evals and Prompts API shut down on 30 November 2026: how to migrate

The OpenAI Agent Builder shutdown is scheduled for 30 November 2026, together with the Evals platform and the v1/prompts API. Existing evals become read-only even earlier, on 31 October 2026. If a chatbot, internal assistant or AI feature in your business was built with any of these, it needs to move to the official replacements within the next five weeks.

This article sums up the dates on OpenAI's Deprecations page, what replaces each product, the migration steps OpenAI recommends, and how to check whether your own systems are affected.

What OpenAI is shutting down and when

According to the OpenAI Deprecations page, read on 26 October 2026, OpenAI notified developers on 3 June 2026 about three deprecations:

• Agent Builder is scheduled to shut down on 30 November 2026. OpenAI notes that ChatKit remains available.
• The Evals platform: existing evals become read-only on 31 October 2026, and the Evals dashboard and API are scheduled to shut down on 30 November 2026. Graders documented for eval workflows are part of this transition.
• Reusable prompts: the v1/prompts API and reusable prompt objects are scheduled to shut down on 30 November 2026. Prompt creation in the platform has been de-emphasized since June.

What replaces Agent Builder, Evals and reusable prompts

OpenAI publishes an official migration path for each product:

• Agent Builder: the migration guide points to the Agents SDK, or to ChatGPT Workspace Agents for teams with a ChatGPT Business, Enterprise or Edu workspace. A workflow exported from Agent Builder can be pasted into a new workspace agent to convert it.
• Evals: OpenAI links to a cookbook guide on moving from OpenAI Evals to Promptfoo, an open-source evaluation tool.
• Reusable prompts: the prompt migration guide says to store the prompt text in your codebase and pass the generated messages as input in the Responses API call.

How to migrate prompt objects into your code

OpenAI's guide lists five changes. We added the last step, because a migration that is not tested before the deadline is not finished.

1. 1Move prompt content into source code, so prompt changes go through the same review and release process as product logic.
2. 2Replace prompt variables with function arguments, so dynamic values are explicit and typed in your application.
3. 3Pass messages through input in the Responses API call instead of using the prompt object.
4. 4Move versioning to your repo using Git commits, PR review, and tests or evals.
5. 5Keep static content first and dynamic content later to preserve prompt caching benefits, since cache hits depend on exact prefix matches.
6. 6Test on staging before 30 November, comparing answers from the old prompt object and the new code path on the same inputs.

How to find out if you are affected

Search your codebase and configuration for calls to v1/prompts, for a prompt parameter in Responses API requests, and for workflows created in Agent Builder. Check scheduled jobs and internal tools too: AI features built quickly for a pilot are the ones most often forgotten. If you used Evals to track quality, export what you need before 31 October, while the evals can still be edited.

What it means for your business

Many companies started their AI features with no-code tools like Agent Builder or saved prompts in the dashboard, because it was the fastest way to test an idea. That speed now has a deadline. After 30 November, any request that depends on these products will fail, and a customer-facing chatbot will simply stop answering.

The good news is that the replacement is a better place for prompts to live. Prompts in code are reviewed, versioned and tested like the rest of your product, which makes AI features easier to maintain. If you are still choosing a provider, our guide on how to choose an AI model for your business explains what to compare beyond the price per token.

How Estontinc helps

Estontinc is a software house in Malang, Indonesia, with 100+ projects and 50+ clients over more than six years. For OpenAI integrations we can:

• Audit your AI integrations for Agent Builder workflows, Evals usage and prompt objects, and deliver a list before the deadline.
• Migrate and test prompts into your codebase and flows to the Agents SDK, as part of our AI development service.
• Build the surrounding application with our custom software team: APIs, admin panels and monitoring.
• Keep it maintained after release with bug fixes, updates and monitoring when providers change their APIs again.

We work under NDA, you own the source code, and we communicate in English, Indonesian or Japanese. Want a quick review of your OpenAI setup before 30 November? Send the details through the contact page and we will reply within 24 hours.