Custom GPT creation is already blocked on Free, Go, Plus, and Pro accounts, and existing GPTs are scheduled to retire on Dec 11, 2026. If a GPT is part of your workflow, the reusable parts carry over into the replacement, plugins, and the rebuild takes a chat, not code.
What a custom GPT is
A custom GPT is a version of ChatGPT configured for a specific purpose. It bundles your own instructions, uploaded knowledge files, selected capabilities like web search or image generation, and conversation starters. An editor GPT carries your style guide and rewrites to it. A planning GPT works from your own checklists. A link or the Explore GPTs menu opens it, and every chat starts with your setup instead of a blank page.
What's happening, and when
Creation stopped first: personal ChatGPT accounts (free, Go, Plus, and Pro) can no longer create or publish new GPTs. Business, Enterprise, and Edu workspaces still allow it where their settings do, and OpenAI's planned freeze for Enterprise creation is Oct 26, 2026.
Retirement follows. Custom GPTs and their pages become inaccessible on Dec 11, 2026, though Enterprise workspaces with an approved deferral get until Feb 11, 2027. Until then, existing GPTs keep working, and the conversations you've had with them remain available after retirement.
Plugins, the replacement
A plugin is a reusable ChatGPT setup built from three parts. A skill holds the instructions, reference files hold the documents it should use, and apps connect it to outside services. For a simple writing assistant, the skill might say "rewrite this in my structure and tone," while the reference files contain your style guide. You only need connected apps when the workflow depends on another service.
The plugin directory lives in the ChatGPT sidebar on web, desktop, and mobile, and it's available across plans, though access to an individual plugin depends on your plan, workspace, role, and region. ChatGPT can also use an installed plugin on its own when its description matches your request, or you can call one with an @ mention.
How to build one without coding
OpenAI's Plugin Creator does the building through conversation. Install it from the plugin directory, then start a chat with @plugin-creator and describe the job. OpenAI's own example: a plugin that gathers progress from Slack and Linear and drafts a weekly update with milestones, blockers, and next steps, following an attached template to keep the format consistent. You test it on a task from your own work, mention @plugin-creator again to edit, and add app connections if the plugin needs them.
If you haven't built a GPT, we'd start with what the free plan already includes: Custom Instructions stores preferences that apply across your chats, a Project keeps instructions and files together for one kind of work, and scheduled tasks rerun a saved prompt when a task repeats. Our automation walkthrough shows that path.
Migrating a GPT you made
The rebuild gets easier once you split a GPT into its parts: which pieces were instructions, which were files, and which depended on an outside connection. Migration follows that split. When it's available for your account, My GPTs shows a Migrate to plugin option, and instructions become a skill, knowledge files become reference files, and connected apps become apps.
Three things don't carry: custom actions (API connections that need rebuilding, sometimes with a custom server), the selected model, and sharing settings, since a migrated plugin starts private. The original GPT stays usable until retirement but turns read-only after migration, and only its latest published version transfers, so finish essential edits before you move it.
The testing habit we'd keep: run the prompts you know plus one harder case, and check that the plugin picks the right skill, uses the expected files, and produces the format you need.
If you use someone else's GPT
Using someone else's GPT doesn't guarantee access to its replacement. The notice on the GPT page is the first place to look, along with any guidance its creator published. Permission to use a GPT doesn't include permission to migrate it, so if the creator built a replacement plugin, check that it's available to your account before you plan around it.
FAQ
Can I still create a custom GPT?
Not on a personal account. Creation and publishing are blocked on free, Go, Plus, and Pro plans. Business, Enterprise, and Edu workspaces can still create GPTs where their settings allow, with new creation planned to end Oct 26, 2026.
When do custom GPTs stop working?
Dec 11, 2026 is the scheduled retirement date, when GPTs and their pages become inaccessible. Enterprise workspaces with an approved deferral get until Feb 11, 2027.
What happens to my past chats with a GPT?
They stay accessible after retirement. OpenAI says no action is needed to preserve them.
What is a ChatGPT plugin?
A bundle of skills (reusable instructions), reference files, and connected apps that works alongside your chats. Some plugins can link to outside services or respond to events, like a document changing or a new reply in a thread.
Do I need a paid plan to use plugins?
The plugin directory is available across ChatGPT plans. Using an individual plugin depends on your plan, workspace, role, region, and what the plugin includes.
What happens to custom actions in a GPT?
They don't transfer through migration. Check whether an available app covers the tasks first; rebuilding the connection can take a custom MCP server and technical setup.
What should I save before migrating?
Migration uses the GPT's latest published version, so finish essential edits before migrating, since drafts and unpublished edits don't transfer. Review the instructions, files, and connected apps after the move, then test before relying on it.
Want to go further?
- How to Automate Repetitive Tasks with ChatGPT shows the free-plan path for repeated tasks: saved setups, Projects, and scheduled tasks.
- What Is an AI Workflow? Building Your First One breaks a task into input, instructions, draft, and review, the same parts a plugin skill organizes.