How to Create a Custom GPT (2026): Who Can, Steps, Limits

As of September 16, 2026, you can create a new custom GPT only inside a ChatGPT Business, Enterprise, or Edu workspace that allows it: open Explore GPTs (chatgpt.com/gpts), select Create, add instructions, up to 20 knowledge files of up to 512 MB each, capabilities, and apps or actions, then test in Preview and select Create. Personal accounts, including Free, Go, Plus, and Pro, cannot create or publish new GPTs, and OpenAI is planning to retire custom GPTs and move them to plugins (OpenAI Help Center, September 2026). GPTs you already built keep working until retirement, and you can still edit them if your plan and permissions allow it.

This walkthrough follows OpenAI's current Help Center articles on creating, sharing, and retiring GPTs, re-read on September 16, 2026: who can build, the Create and Configure views, knowledge limits, actions, GPT Store publishing, the retirement timeline, how to migrate a GPT to a plugin, what happens when a subscription ends, and what to use instead on a personal plan.

By Adi Leviim, Co-Founder at Infi Developments. Last reviewed September 16, 2026 against OpenAI's Help Center.

Who Can Create a Custom GPT in 2026?

Only members of a ChatGPT Business, Enterprise, or Edu workspace can create a new custom GPT, and only when the workspace settings and their role allow it. OpenAI's GPTs article states: "New GPT creation and publishing are not available on personal ChatGPT accounts, including Free, Go, Plus, and Pro." Its troubleshooting article adds that this "is an account eligibility restriction, not a technical error," so completing a builder profile or appealing will not unlock it.

Creation and editing happen on the web; OpenAI says the mobile apps support using GPTs but not creating them. One inconsistency to know about: the comparison table on OpenAI's pricing page still shows "Create & share GPTs" as included on Go, Plus, and Pro. The Help Center articles on GPTs, which now carry the retirement notice, are the more specific source, so this guide follows them. ChatGPT Business costs $25 per user per month billed monthly or $20 billed annually for a Standard seat, and a workspace needs at least two seats, according to OpenAI's Business overview. Before paying for a workspace only to build GPTs, keep in mind that new creation is planned to end on October 26, 2026 for affected Enterprise workspaces and that other plans may follow.

What Is a Custom GPT?

A custom GPT is a version of ChatGPT configured for one purpose, combining its own instructions, uploaded knowledge, and selected tools, that anyone with access can open and chat with. OpenAI lists six building blocks:

• Instructions: the GPT's behavior, tone, goals, and boundaries.
• Conversation starters: example prompts shown when someone opens the GPT.
• Knowledge: uploaded files the GPT uses as reference.
• Capabilities: tools such as web search, image generation, Canvas, and Code Interpreter & Data Analysis.
• Apps: tools and services that users have connected.
• Actions: external APIs you define with an OpenAPI schema.

Two details surprise people: GPTs "do not use saved memory, custom instructions, or previous conversations," and builders cannot view individual conversations users have with their GPTs. On ChatGPT web you can also type @ in a normal chat to bring in a GPT (not in the mobile apps).

How to Create a Custom GPT, Step by Step

In an eligible workspace, creating a GPT takes six steps: open the GPT builder, choose the Create or Configure view, write instructions, add knowledge, pick capabilities and integrations, then test in Preview and save. The steps follow OpenAI's "Creating and editing GPTs" article; screenshots show the web builder captured earlier in 2026, and your options depend on workspace settings.

Who these steps are for: Business, Enterprise, and Edu members whose workspace still allows GPT creation, and anyone editing an existing GPT. Creation is planned to end on October 26, 2026 for affected Enterprise workspaces, other plans may follow, and every GPT is scheduled for retirement, so treat a new GPT as a short-term build that you will migrate to a plugin. If you plan to migrate it, publish it before the cutoff: drafts cannot be migrated, and publishing does not require sharing it publicly.

Step 1: Open the GPT builder

Open Explore GPTs in the ChatGPT sidebar or go to chatgpt.com/gpts, then select Create. If Create is missing or greyed out, you are most likely signed in to a personal account or your workspace role does not include GPT creation.

Step 2: Choose Create (conversational) or Configure

The builder opens with two tabs. Create is the conversational builder: you describe what you want and ChatGPT drafts the GPT for you. Configure is the configuration view, where you edit each field directly. Draft in Create, then switch to Configure to tighten the fields: Name (shown in search, the GPT Store, shared links, and the chat header), Description (shown in previews and Store listings), and a few realistic Conversation starters.

Step 3: Write the instructions

Instructions apply to every conversation. OpenAI's guidance: use explicit step structure ("When X happens, do Y"), prefer concrete "Do X" rules over long prohibition lists, add brief examples of acceptable and unacceptable output, and use headings and lists. If the GPT should cite your files, say so and specify the format. Keep a copy of the final instructions outside the builder: in the planned migration, the instructions become a skill in the new plugin, and you will want the original to compare against.

Step 4: Add knowledge files

Under Knowledge, upload reference material such as handbooks, documentation, or product specs. You can attach up to 20 files, and each file can be up to 512 MB. Most document, spreadsheet, image, text, and code types are supported; some types only work when Code Interpreter & Data Analysis is enabled. Put rules and tone in instructions, not files, and prefer clean, text-forward documents; OpenAI notes that complex layouts can make content harder to use and that images inside files may not be reliably interpreted. Keep the source files too, since OpenAI asks you to review migrated reference files before relying on the replacement plugin.

Step 5: Set the recommended model, capabilities, and apps or actions

The Recommended model is the model suggested when a user starts a chat; users can still switch, and if a model is no longer available, OpenAI automatically switches the GPT to a similar current model. The selected model does not carry over when a GPT is migrated to a plugin. Under Capabilities, enable only what the GPT needs; OpenAI's builder article lists Web search, Image generation, Canvas, Code Interpreter & Data Analysis, and Apps. Availability depends on your account, workspace, and region. Below that, Actions connect the GPT to an API you define. A GPT can use apps or actions, but not both at the same time. If you are choosing between them for a GPT you will migrate, note that connected apps are added to the plugin as apps, while custom actions do not transfer.

Step 6: Test in Preview, then save

Use the Preview pane to run real prompts and check tone and accuracy. Changes save to a draft automatically; select Create when you are ready, or Update when editing an existing GPT. Later, open the more options menu (•••) to copy the GPT link, view and restore version history, or delete the GPT. To edit again, go to Explore GPTs, My GPTs, choose the GPT, and select Edit GPT.

Custom GPT Knowledge Limits, Capabilities, and Actions

The limits that matter most are 20 knowledge files at up to 512 MB each, one integration path per GPT (apps or actions), and a valid Privacy Policy URL on every action in a GPT you share publicly. This table collects the numbers and rules from OpenAI's builder, actions, and sharing articles.

In Enterprise and Edu workspaces, admins can restrict which domains actions may call. If the editor shows "No domains are allowed by your workspace's settings," actions cannot run until an admin allows a domain.

How to Share a GPT or Publish It to the GPT Store

To share a GPT, open it in the editor, select Share, choose a sharing level and a permission level, and select Save; publishing to the GPT Store additionally needs account eligibility, workspace permission, and passing automated policy checks. Personal accounts cannot publish new GPTs. At retirement, custom GPTs and their GPT pages are scheduled to become inaccessible, so a Store listing is not a long-term channel; a migrated plugin starts private and is shared through plugin settings instead.

• Sharing levels: Invite-only (only you on a personal account, or chosen users and groups in a workspace), Anyone at your workspace with the link, Your workspace, Anyone with the link, and GPT Store.
• Permission levels: Can chat, Can view settings (use, duplicate, and view configuration), and Can edit. Public link sharing and the GPT Store support only Can chat.
• GPT Store requirements: you may be asked to choose a category, review how your builder name or website appears, and confirm policy compliance. A complete builder profile may be required, and only verified domains can appear as its website link.
• Common blockers: public publishing disabled for the workspace, app connections not supported for that sharing option, actions without a Privacy Policy URL, or a policy restriction. If an appeal is available, the GPT stays usable privately but cannot be edited or shared while the appeal is under review.

What Happens to Custom GPTs If Your Subscription Ends?

If you cancel or downgrade, you can keep using the GPTs you created, but editing them depends on your new plan and permissions, and no personal plan can create new GPTs. Both points only hold until retirement, covered next.

Are Custom GPTs Being Retired? Timeline and What Changes

Yes. OpenAI says it is "planning to retire custom GPTs" and recommends moving workflows to plugins, which package reusable instructions (as skills) with connected apps. The announcement appeared in ChatGPT's release notes on September 11, 2026, and OpenAI's retirement FAQ gives these planned milestones for affected Enterprise workspaces:

OpenAI says the transition affects all ChatGPT plans and that other plans "may follow the same transition timeline," with in-product announcements; it does not guarantee that migration reaches every account or workspace on the target date, and the dates are subject to change. Existing GPTs remain usable until retirement. Public GPTs created in affected Enterprise workspaces are included even when you use them from another plan, and access to a public GPT does not guarantee access to its replacement plugin. What the planned migration does and does not carry over:

• Transfers: the GPT's instructions become a skill in the new plugin, and connected apps are added as apps.
• Does not transfer: custom actions (they must be rebuilt with a supported connector or custom MCP server) and the selected model (in Enterprise workspaces, Enterprise defaults apply). Conversation starters and previous chats may not copy.
• After migration: the original GPT stays usable until retirement but becomes read-only, and its creator cannot delete it. In the planned Enterprise workflow, only published GPTs can be migrated, and only by the GPT's creator or a workspace admin.
• At retirement: links to migrated GPTs are planned to redirect to the replacement for people who have access; the redirect does not grant access.

How to Migrate a Custom GPT to a Plugin

Migration is a built-in button on a published GPT: select it when your account or workspace offers it, then review, test, and share the plugin it creates. For Enterprise workspaces, the migration experience is targeted for September 22, 2026; personal accounts should follow the instructions in their in-product notice. These steps follow OpenAI's retirement and migration FAQ:

1. Take inventory. List the GPTs you rely on, who created each one, and who needs access to the replacement. Record each GPT's instructions, reference material, integrations, and a few familiar prompts to test with later. In an Enterprise workspace, coordinate with your admin.
2. Finish edits and publish. Make essential edits first, because the original becomes read-only after migration. Drafts cannot be migrated directly, so publish any draft you need; publishing does not require sharing it publicly.
3. Check who can migrate. In the planned Enterprise workflow, only the GPT's creator or a workspace admin can migrate it, plugins must be enabled, and the GPT must be published. Permission to use someone else's GPT does not let you migrate it. The built-in migration does not require the Upload plugins permissions.
4. Start the migration. Open your published GPT and select its migration button, or ask a workspace admin for help. If you do not see the option yet, it may not have reached your account, or plugin access may be disabled.
5. Review what transferred. Check that the instructions arrived as a skill and connected apps as apps, then review the migrated reference files, templates, examples, and tools. Re-add conversation starters or context you still need, since they may not copy.
6. Rebuild custom actions. Actions do not transfer. Rebuild required integrations with a supported connector or a custom MCP server, review permissions with your technical and security teams, and test before switching. A rebuilt integration may not provide every capability of the original action.
7. Test against the original. Run familiar prompts and at least one harder case. Confirm the plugin selects the right skill and follows your instructions, uses the expected reference material, produces complete answers or files in the required format, and has the tools and integrations the workflow needs.
8. Share the plugin. The replacement plugin starts private. Review its sharing settings, share it, and ask an intended user to install and test it. In Enterprise workspaces, sharing with people or groups requires the Share plugins permission, and publishing to the workspace directory requires Publish plugins to workspace.
9. Maintain the plugin from now on. Make future changes in the plugin, not the GPT. Users open it with an @ mention or from + then More, where supported, and ChatGPT may also use an installed skill automatically when its description matches the request.

Users of the migrated plugin need access to it and an account where plugins are available; in an Enterprise workspace they also need the Use plugins permission and a policy that lets them install the plugin or installs it for them. Any included apps still need their own access and authorization.

Keep the work behind your GPTs before the migration. AI Toolbox (formerly ChatGPT Toolbox) saves your GPT instructions as reusable prompts you insert with //, exports the chats you had with a GPT, and files GPTs, chats, and projects into the same folders. Free plan available; 40,000+ users, rated 4.7/5 from 800+ ratings on the Chrome Web Store. Install AI Toolbox free ->

Can't Create a GPT on Plus or Pro? What to Use Instead

On a personal plan, the closest native replacement is a ChatGPT project, which combines project instructions, reference files, and chats in one workspace and is available on every plan, including Free. Project instructions apply only inside that project and override your global custom instructions. OpenAI lists project file limits of up to 5 files on Free, 25 on Go and Plus, and 40 on Pro. See our guide to using ChatGPT projects for the setup.

• Plugins and skills: OpenAI's pricing table lists Plugins and a Skills beta on personal plans, and plugins are where OpenAI recommends moving GPT workflows. OpenAI's plugins article says installing or using a plugin depends on your plan, workspace, role, region, and the plugin's capabilities, so check your own account.
• Custom instructions: for one style you want everywhere, account-level custom instructions are simpler than a GPT; our custom instructions templates give starting points.
• Gemini Gems: if you use both assistants, Google's Gems help page lists an age requirement for personal Google Accounts but no paid plan requirement; see how to create a Gemini Gem. Like custom GPTs, Gems are being retired: Google is replacing them with skills, starting November 2026 for personal accounts.

If you already have a workspace and want deeper configuration advice for the GPTs you still maintain, the custom GPT builder guide goes further into instruction design.

How AI Toolbox Helps You Manage GPTs and Their Chats

AI Toolbox is a Chrome extension that does not create or migrate GPTs; it keeps the material around them organized, searchable, and exportable inside ChatGPT on any plan. That matters now because OpenAI says previous chats may not copy into a migrated plugin and custom actions have to be rebuilt.

Folders use search-to-add: open a folder and add chats, GPTs, or projects by searching for them. The export and search walkthroughs are in how to export ChatGPT conversations, how to export ChatGPT projects, building a prompt library with the // shortcut, and finding old ChatGPT chats. Teams that need this across seats can look at AI Toolbox Teams, which offers a 14-day free trial.

Who Is Affected, and What to Do This Month

Consultants and agencies with client GPTs on Plus or Pro can no longer duplicate them for new clients, since duplication requires eligibility to create. Save each client's instructions and conversations now.

Enterprise admins should inventory published GPTs, their creators, and custom actions; drafts cannot be migrated, September 17 is the target for the migration experience, and September 25 is the planned end of new creation.

Developers who built GPT actions should start rebuilding the integrations they need as a connector or custom MCP server now, since actions do not transfer and a rebuilt integration may not match every capability of the original.

Teachers and trainers who share a GPT by link should plan a project or plugin replacement: migrated links are planned to redirect only for people who have access to the replacement.

Everyone who uses someone else's GPT should check any notice on the GPT and guidance from its creator, because only the creator or a workspace admin can migrate it.

Limitations and What Native ChatGPT Does Well

Only OpenAI's own tools can create, edit, share, or migrate a GPT, and nothing in AI Toolbox changes those rules. The native builder does its job well: the Create view drafts a GPT conversationally, Preview makes testing fast, version history lets you roll back, and projects cover much of what personal users built GPTs for. What OpenAI has not published: confirmed retirement dates for plans other than Enterprise, detailed migration steps outside Enterprise, and whether the pricing table will be corrected. Watch your account's in-product notice and treat the Enterprise dates as a signal, not a promise.

Frequently Asked Questions

Can I still create a custom GPT with ChatGPT Plus?

No. OpenAI's Help Center states that new GPT creation and publishing are not available on personal ChatGPT accounts, including Free, Go, Plus, and Pro. You can still use GPTs and edit GPTs you already created if your plan and permissions allow it. New GPTs can be created only in Business, Enterprise, and Edu workspaces that permit it, and only until creation ends as part of the custom GPT retirement.

How do I create a custom GPT?

In a Business, Enterprise, or Edu workspace that still allows it, open Explore GPTs or chatgpt.com/gpts and select Create. Draft the GPT in the conversational Create view or fill in the Configure view: name, description, instructions, knowledge files, capabilities, and apps or actions. Test it in Preview, then select Create to save. For affected Enterprise workspaces, creating new GPTs is planned to end October 26, 2026, so publish anything you want to migrate before then.

Are custom GPTs being discontinued?

Yes. OpenAI announced on September 11, 2026 that it is planning to retire custom GPTs across ChatGPT plans and recommends moving workflows to plugins, with a migration path planned. Existing GPTs remain usable until retirement. OpenAI says the transition affects all ChatGPT plans, other plans may follow the Enterprise timeline, and the dates are subject to change.

When do custom GPTs stop working?

For affected Enterprise workspaces, retirement is scheduled for December 11, 2026, when custom GPTs and their GPT pages become inaccessible. Before that, the migration experience is targeted for September 22, 2026 and creation of new custom GPTs is planned to end October 26, 2026. Other plans may follow the same timeline; watch the in-product notice for your account or workspace.

How do I migrate my GPT to a plugin?

When migration is available to you, open your published GPT and select its migration button, or ask a workspace admin for help; personal accounts should follow their in-product notice. In the planned Enterprise workflow, only published GPTs can be migrated, by their creator or a workspace admin, and plugins must be enabled. The instructions become a skill and connected apps become apps. Then review the plugin, test it with familiar prompts, and share it, because the plugin starts private and the original GPT becomes read-only.

What happens to GPT actions?

Custom actions do not transfer through the migration workflow. Whoever maintains the workflow needs to rebuild the required integrations with a supported connector or a custom MCP server, review permissions with technical and security teams, and test before switching. OpenAI warns that a rebuilt integration may not provide every capability of the original action.

What is a custom GPT?

A custom GPT is a version of ChatGPT configured for a specific purpose. It combines its own instructions, conversation starters, uploaded knowledge files, and selected capabilities such as web search or image generation, and it can connect to outside services through apps or actions. GPTs do not use your saved memory or custom instructions.

How many files can a custom GPT have?

You can attach up to 20 knowledge files to a GPT, and each file can be up to 512 MB. Most document, spreadsheet, image, text, and code formats are accepted, and some file types work only when Code Interpreter and Data Analysis is enabled for the GPT.

What happens to my custom GPTs if I cancel my subscription?

You can keep using the GPTs you created. Editing them depends on your subscription and permissions, so moving to a plan that does not support editing may stop you from making changes. Creating new GPTs is unavailable on every personal plan, and existing GPTs are separately scheduled for retirement.

How do I publish a GPT to the GPT Store?

Open the GPT in the editor, select Share, and choose GPT Store if it is available to you. You may need to pick a category, complete a builder profile, and pass automated policy checks, and every public action needs a valid Privacy Policy URL. Personal accounts cannot publish new GPTs, workspaces must permit public publishing, and custom GPTs and their GPT pages are scheduled to become inaccessible at retirement.

What can I use instead of a custom GPT on a personal plan?

A ChatGPT project is the closest native option: it holds project instructions, reference files, and related chats, and it is available on every plan. OpenAI lists up to 5 project files on Free, 25 on Go and Plus, and 40 on Pro. OpenAI's pricing page also lists plugins on personal plans, and plugins are where OpenAI recommends moving GPT workflows, although availability depends on your plan, region, and the plugin.

Bottom Line

Creating a custom GPT in 2026 means working inside a Business, Enterprise, or Edu workspace that still allows it: open Explore GPTs, select Create, configure instructions, up to 20 knowledge files, capabilities, and apps or actions, then test and save. On Free, Go, Plus, or Pro you can use and, where allowed, edit existing GPTs, but not build new ones. Every custom GPT is on a path to retirement, scheduled for December 11, 2026 in affected Enterprise workspaces, so the durable move is to inventory your GPTs, save their instructions and the conversations you depend on, migrate to a plugin when it is offered, and rebuild any custom actions.

Related reading: how to cancel a ChatGPT subscription, ChatGPT pricing in 2026, and ChatGPT folders.

Sources

• OpenAI Help Center, GPTs in ChatGPT
• OpenAI Help Center, Creating and editing GPTs
• OpenAI Help Center, Sharing and publishing GPTs
• OpenAI Help Center, Custom GPT retirement and migration FAQ
• OpenAI Help Center, Plugins in ChatGPT and Codex
• OpenAI Help Center, ChatGPT release notes (September 11, 2026: Planned custom GPT retirement and migration to plugins)
• OpenAI Help Center, Configuring actions in GPTs
• OpenAI Help Center, Troubleshooting GPTs
• OpenAI Help Center, Projects in ChatGPT
• OpenAI Help Center, ChatGPT Business overview
• OpenAI, ChatGPT pricing
• Google Gemini Apps Help, Use Gems in Gemini Apps

Last updated: October 7, 2026