# OpenAI Evals, Agent Builder and v1/prompts Shut Down on 30 November

Per OpenAI's deprecations page, existing evals became read-only on 31 October 2026, and the Evals dashboard and API, Agent Builder and the v1/prompts API all shut down on 30 November. Announced on 3 June, the shutdowns are now 27 days away.

By Neomanex. Updated 2026-11-03.

**Our verdict: treat 30 November as a hard outage date and migrate this week.** OpenAI Evals, OpenAI Agent Builder and the v1/prompts API (reusable prompt objects) are all scheduled to shut down on 30 November 2026, per OpenAI's own deprecations page. Evals already passed its first milestone: on OpenAI's schedule, existing evals became read-only on 31 October. As of 3 November, that leaves 27 days. None of this is new (OpenAI announced all three retirements on 3 June), but we had not covered it, and the calendar is now the story.

## What happened

OpenAI's deprecations page lists three retirements, each announced on 3 June 2026. We re-read the page on 3 November: none of the dates have moved.

### Evals: read-only since 31 October, shut down on 30 November

"On June 3, 2026, we notified developers using the Evals platform that the product is being deprecated." The schedule:

- **31 October 2026:** "Existing evals become read-only."
- **30 November 2026:** "The Evals dashboard and API are scheduled to shut down."

"Graders documented for eval workflows are part of this transition." OpenAI's migration guidance points to Promptfoo, an open-source evaluation tool.

### Agent Builder: shut down on 30 November

"Agent Builder is scheduled to shut down" on 30 November 2026. OpenAI's migration path is to continue with "the Agents SDK or ChatGPT Workspace Agents". ChatKit remains available, so a ChatKit front end can stay while the agent behind it moves.

### Reusable prompts and v1/prompts: shut down on 30 November

"The v1/prompts API and reusable prompt objects are scheduled to shut down" on 30 November 2026. OpenAI's instruction is to move reusable prompt content into your application code.

## Why it matters

These are three separate failure points with one shared date. If your CI calls the Evals API to gate releases, that step breaks on 30 November. If an agent was built in Agent Builder, it stops. If any request references a stored prompt object by id, it fails unless the prompt text has been inlined or moved into your own config. And because evals are already read-only, you can no longer edit an eval to buy time: the only remaining move is to copy each eval and recreate it elsewhere.

## What changes for you

- **Evals:** copy every eval and grader you still need now, then recreate the runs in another tool. OpenAI's own guide points to Promptfoo and rebuilds evals by hand, without an export feature.
- **Agent Builder:** move each agent to the Agents SDK or ChatGPT Workspace Agents. Keep ChatKit if you use it.
- **Prompts:** search your codebase for v1/prompts calls and prompt object ids, and move that prompt text into application code.
- **Deadline:** all three must be done before 30 November 2026.

## FAQ

**Can I still edit an existing eval?**
No. On OpenAI's schedule, existing evals became read-only on 31 October 2026. You can still read them until the 30 November shutdown.

**Does ChatKit shut down too?**
No. OpenAI's deprecations page says ChatKit remains available. Only Agent Builder is retired.

**What does OpenAI recommend instead of Evals?**
OpenAI's migration guidance points to Promptfoo, an open-source evaluation tool.

Canonical: https://neomanex.com/news/openai-evals-read-only-oct-31-agent-builder-prompts-api-shut-down-nov-30