SunsettingAI Dev Tools

OpenAI Reusable Prompts (prompt objects and the v1/prompts API)

by OpenAI

Announced

2026-06-03

Complete

2026-11-30

Refund window closes

—

Refund status

No refund applicable

On June 3, 2026, OpenAI notified developers using reusable prompts in the dashboard and API that reusable prompt objects are being deprecated, and prompt creation was de-emphasized in the platform. Per OpenAI's deprecations page, the v1/prompts API and reusable prompt objects are scheduled to shut down on November 30, 2026. Today a Responses API call can reference a saved prompt with a prompt object holding its ID (pmpt_...), an optional version and variables. OpenAI's migration guide says to move the prompt content into your application code and pass the messages directly as input instead. The guide does not describe a bulk export tool for saved prompts.

Refund flow

1
Not applicable. OpenAI's deprecation notice and migration guide describe no refund or credit for this change.
2
Before November 30, 2026: for each saved prompt your code references, copy its full text, its variable names, and the version your code pins. The migration guide does not describe a bulk export tool, so do this per prompt.
3
Search your codebase for every request that sends a prompt object with a pmpt_ ID, so you know which calls must change.
4
Rewrite each of those calls to send the messages through input in the Responses API, and test them before the shutdown date rather than on it.

Migration path

Recommended successor

Prompts in your own code, passed as input to the Responses API

Why this is the closest fit

It is the path OpenAI documents. You still call the Responses API, but you send the messages yourself through input instead of pointing at a saved prompt ID. OpenAI says this gives you more control over review, testing, deployment and versioning.

What differs from the original

Per OpenAI's guide, there is no managed prompt object any more. Variables become function arguments, so dynamic values are explicit in your application. Versioning moves to your repo, through git commits, PR review, and tests or evals. The guide also says to keep static content first and dynamic content later to preserve prompt caching, since cache hits depend on exact prefix matches. Instead of scattering prompts inline, OpenAI recommends a small prompts/ module with one named builder function per prompt, plus lightweight eval fixtures so prompt changes are reviewed like product logic.

Alternatives by use case

You call a handful of saved prompts from one app

Move each into a named builder function in a prompts/ module and pass its output as input, as OpenAI's guide shows

Your saved prompts use variables

Turn each variable into a function argument, and keep the fixed instructions at the start of the message list so prompt caching still has an exact prefix to match

You relied on dashboard versions to roll prompts forward or back

Version prompts in git with PR review, and manage rollout through your own config or feature flags, which the guide lists as an option

You use Codex

OpenAI's guide points to its Developers plugin and OpenAI Docs skill as a way to automate the migration. Review the diff before you ship it

What this tool meant

Reusable prompts let teams keep prompt text in the OpenAI dashboard and call it by ID, with versions and variables, so a prompt could change without a code deploy. Its shutdown is part of a June 3, 2026 batch in which OpenAI also scheduled Agent Builder and the Evals platform for shutdown on the same November 30 date. For anyone generating AI video prompts through an API, the prompt is the asset: the wording, the variables for subject, style and camera, and the version that produced the clips you liked. If that wording only exists in a vendor dashboard, a shutdown notice puts it at risk. Keeping prompts in your own files, with history, means a change of tool costs a migration, not a rewrite from memory.