Keep Custom GPTs Available in ChatGPT

Earlier this month, OpenAI announced that Custom GPTs will stop running on December 11, 2026, as the company moves toward Plugins.

We are asking OpenAI to reconsider that decision and keep Custom GPTs available alongside Plugins.

Custom GPTs are more than saved prompts. They let users create stable, purpose-built environments with persistent instructions, uploaded knowledge and defined workflows, while still allowing each new chat to begin cleanly. That separation matters. In many professional use cases, prior conversations should not influence the next one. When memory carries assumptions, facts or conclusions from one chat into another, it can distort the analysis and reduce the quality of the results.

For many users, the value of a Custom GPT is exactly that it can be configured once and then used repeatedly without rebuilding the same context every time.

Projects are not a replacement for Custom GPTs. Projects can be useful for ongoing work where context should carry across conversations. But that is a different use case. In many professional workflows, each chat must be handled in isolation. Prior conversations can introduce assumptions, facts, preferences or conclusions that do not belong in the current task. When memory or project context bleeds across chats, it can distort the output and reduce the reliability of the results.

Custom GPTs allow users to preserve the same instructions and knowledge while starting each conversation from a clean context. That separation is essential for work such as document review, proposal analysis, evaluation, research, compliance, due diligence and other tasks where each case should be assessed independently. For these users, memory bleed is not a minor inconvenience. It can undermine the quality of the work.

OpenAI's own migration guidance acknowledges that a Plugin created from an existing GPT may behave differently from the GPT it replaces. That means migration does not necessarily preserve what users have built.

Plugins can also require organizations to enable and govern connectors to external systems. That creates an entirely different deployment model from a Custom GPT built around fixed instructions and uploaded knowledge. Organizations may have to approve access, configure permissions, manage roles and evaluate security implications simply to recreate workflows that previously did not require a live connection to enterprise systems.

Thousands of individuals and organizations have invested significant time creating, testing and refining Custom GPTs. Some have built entire workflows around them. Removing the feature risks eliminating significant accumulated work and a form of customization that many users specifically chose ChatGPT for.

We support OpenAI continuing to develop Plugins and new ways of extending ChatGPT. We are asking that this development expand what users can build rather than require the elimination of Custom GPTs. We ask OpenAI to:

• Keep existing Custom GPTs operational after December 11, 2026.
• Continue allowing users to create and edit Custom GPTs, even if Plugins become the primary development platform.
• Preserve the ability to start clean, isolated conversations that use the same instructions and knowledge without carrying memory or context from unrelated chats.
• If maintaining the current system indefinitely is not possible, provide a migration path that preserves Custom GPT instructions, knowledge, behavior and functionality as closely as possible.
• Give users sufficient time and tools to migrate complex GPTs before ending support.
• People invested in Custom GPTs because OpenAI gave them a powerful way to shape ChatGPT around their own work, expertise and ideas.

Please preserve that capability.

OpenAI: Keep Custom GPTs available alongside Plugins.