Gemini Gems Equivalent in ChatGPT: Matching Reusable Setups
ChatGPT and Gemini both let you save a custom assistant with standing instructions. This guide explains how a Gemini Gem maps to the closest ChatGPT option and what to keep in mind when you run the same workflow on both.
Updated 5 min read
The closest ChatGPT equivalent to a Gemini Gem is a ChatGPT Project with custom instructions: both give a persistent assistant a name, a set of standing instructions and optional reference material before each conversation begins. Tools like fenbs let you keep the source instructions and project rules outside either app, so the same setup is available regardless of which assistant you are using that day.
The two products use different names and live in different places in the interface, but the underlying idea is the same: write instructions once, skip re-explaining them every session. The differences matter mainly when you want to share the assistant with others or attach reference files.
What a Gemini Gem does
A Gem is a saved configuration inside Gemini (opens in a new tab). You write system instructions that describe how the assistant should behave, what it should focus on, and what tone to use. You can also attach knowledge files the assistant draws on. Every new conversation with that Gem starts from those instructions automatically.
Google Workspace accounts can create Gems for a team, making the same configured assistant available to colleagues. Personal accounts can create Gems for individual use. From November 2026, Google is renaming Gems to skills on personal accounts; check Google's Gemini help center (opens in a new tab) for the current name in your interface.
Related reading: what Gemini Gems are and how they work, and how they compare to Claude Projects vs Gemini Gems.
The closest ChatGPT setup today
ChatGPT Projects are the main way to create a persistent assistant with instructions in ChatGPT. A Project holds a name, custom instructions and uploaded files. Any conversation started inside that Project picks up those instructions from the beginning.
OpenAI's Custom GPTs served a similar role and allowed a public sharing URL. OpenAI has announced Custom GPTs will retire on December 11, 2026; check OpenAI's help center (opens in a new tab) for the current status and any replacement options. For most everyday use, a Project with custom instructions covers the same ground.
ChatGPT also has a Custom instructions setting that applies across all conversations rather than inside one Project. That is useful for personal preferences ("always answer in plain English; I am not a developer") but is not a per-task equivalent of a Gem.
How the two compare
| Capability | Gemini Gems | ChatGPT Projects |
|---|---|---|
| Standing instructions | Yes, written in the Gem setup | Yes, written in Project instructions |
| Knowledge files | Yes, attach documents the Gem references | Yes, upload files to the Project |
| Multiple assistants | Yes, one Gem per task or role | Yes, one Project per task or role |
| Team sharing | Yes, on Google Workspace accounts | Sharing depends on plan; check OpenAI's help for current options |
| Public URL | No direct equivalent | Was available via Custom GPTs (retiring Dec 11, 2026) |
| Works with fenbs over MCP | No direct MCP connection | No direct MCP connection; connect Claude, Cursor or other supported apps instead |
Neither product lets you export instructions directly to the other. You write them in one place, and if you want them in the other, you copy them manually. That copying step is where instructions drift apart over time.
What changes when you move a workflow between the two
The instruction text often transfers without changes. Tone, scope and task description read the same in either interface. Knowledge files need to be re-uploaded because the two products store them separately.
The bigger gap is context about the current work. A Gem or a Project only knows what you gave it during setup, plus whatever you say in the conversation. If a workflow depends on knowing the current project status, a past decision, or what was done in a previous session, that information is not in either app unless you type it in each time.
That is the practical reason to keep a shared record alongside both apps. See the Gemini vs ChatGPT comparison for a broader look at when to use each.
Keeping prompts and project context portable
The instruction text for a Gem or a Project is short enough to copy by hand. The project context it should act on, including decisions your team has made and where things stand, is harder to copy because it changes constantly.
One way to solve this is to write the stable instructions in fenbs as a Skill, and keep project decisions and current task status in fenbs as well. A connected AI app reads both at the start of a session, so the Gem or Project instructions you paste in can stay short: they set the role, and fenbs provides the current facts.
Keep your best instructions in one place
fenbs lets you store tested prompts and Gem instructions as Skills that any connected AI app can read. Add the standing instructions you use in Gemini, note which project they belong to, and they are available when you move to a different assistant. Create a free account to start a Skills library for your regular workflows.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
How to do this with fenbs
- Create a free account and open your personal kanban board.
- Open Skills and prompts. Paste in the instruction text from your Gem or ChatGPT Project. Give it the name of the task it covers, such as "Weekly report editor" or "Code review helper".
- Add the current project to your board. Write the goal, add active tasks, and record any decisions that affect how the assistant should act.
- Connect a supported AI app via Integrations so it can read your Skills and the project context at the start of each session.
- When you switch to a different app, point it at the same fenbs project. The instructions and the context travel with it.
fenbs does not connect directly to Gemini or ChatGPT over MCP; it connects to Claude, Claude Code, Cursor, Codex CLI, Gemini CLI, GitHub Copilot and Devin Desktop. For Gemini and ChatGPT, you paste the relevant context in manually or use a supported app for the session. See the ai-prompts topic hub for more guides on managing instructions across tools.