Can ChatGPT and Claude share memory?

Not on their own. Each keeps its own memory. You can copy ChatGPT's memory into Claude once with Claude's import tool, paste the same context file into both, or connect both to one memory layer such as Membase so they read the same memory.

How do I move my ChatGPT memory to Claude?

In Claude, go to Settings › Your Claude › Start import. Paste Claude's prompt into ChatGPT, copy what ChatGPT writes, paste it back into Claude and select Add to memory.

Does Claude's memory import keep syncing?

No. It's a one-time copy. Anything you tell ChatGPT afterwards won't reach Claude unless you import again.

Can I move Claude's memory to ChatGPT?

Claude's import tool only works into Claude. To go the other way, ask Claude to write out its memories of you and paste them into a ChatGPT chat or its custom instructions, or connect both to a shared memory layer.

Is there a free way to share memory between ChatGPT and Claude?

Yes. Claude's memory import and a context file are free. Membase is also free during its beta.

Can ChatGPT or Claude change my Membase memory?

Not when connected through the consent screen. That connection is read-only and limited to the Memories you tick. Your chats reach Membase through the browser extension.

Does this work with Gemini and Grok too?

Yes, with a shared memory layer. Membase's browser extension captures chats from ChatGPT, Claude, Gemini, Grok, DeepSeek and Kimi, and Grok can also connect to Membase over MCP.

How to Share Memory Between ChatGPT and Claude (2026)

ChatGPT and Claude each keep their own memory, and the two don't sync. Tell ChatGPT about your project and Claude still starts from zero. There are three ways to fix that:

Claude's memory import: copy what ChatGPT remembers into Claude, once.
A context file: keep one document about you and paste it into both.
A shared memory layer: both assistants read the same memory, and it stays up to date.

Short answer: if you're moving from ChatGPT to Claude, use Claude's import. If you use both every day, use a shared memory layer. Options 1 and 2 are free and need no extra tools; option 3 uses Membase, our product.

Option 1: Claude's memory import (one-time copy)

Claude can import memory from another assistant. It's free on every Claude plan, on the web and in Claude Desktop.

In Claude, go to Settings › Your Claude and select Start import. Claude shows a prompt to use.
Paste that prompt into ChatGPT. It writes out what it remembers about you in one block. Remove anything you don't want to move.
Copy the result, paste it into Claude's import box and select Add to memory.
Check what Claude kept with See what Claude learned about you.

Good for: switching from ChatGPT to Claude.

Know that: it's a one-time copy, so anything you tell ChatGPT later won't reach Claude. Anthropic labels memory import as experimental, and Claude's memory focuses on work topics, so some personal details may not be kept. The import only works into Claude, not from Claude into ChatGPT. (Claude Help Center)

Option 2: A context file you paste into both

Write one document with what you want both assistants to know: your role, current projects, preferences, decisions. Paste it at the start of a chat, or into each assistant's custom instructions or project settings.

Good for: a small amount of stable context, at no cost.

Know that: you maintain it by hand. When a decision changes in a ChatGPT chat, you have to update the file and paste it again in Claude.

Option 3: One memory both assistants read

A memory layer keeps your context in one place outside both assistants. ChatGPT and Claude both search it when you ask, so what you add once is available in both, and it keeps up as you work.

The steps below use Membase. It connects to ChatGPT and Claude through MCP, a standard way for AI apps to use outside tools.

Step 1: Create a Memory

Sign up at app.membase.com with Google, X, email or a wallet. Open Memory, select Create memory and give it a name, such as Work. Under the instructions, say what it should keep, or leave Anything useful.

Step 2: Add what you want both assistants to know

On the Memory's page, use Add to this memory:

Your chats: install the browser extension, listed in the Chrome Web Store as Unibase Memory. It captures your conversations in ChatGPT, Claude, Gemini, Grok, DeepSeek and Kimi and hands them to your Memory. Then, on the Memory's page, choose Unibase memory to connect it.
Your notes and files: upload Markdown, text or PDF files, or connect Notion pages.

Then select Update now. Membase learns from what you added, and Your memory on the page shows what it kept. You can edit or forget any entry.

Tip: for a Memory over your chats, write the instructions as a filter, not a topic. "What bears on the Atlas project" keeps what you asked, decided and adopted about it; written as a topic, it tends to produce a log of what each assistant said.

Step 3: Connect Claude

Works on Claude Free (one custom connector), Pro, Max, Team and Enterprise. On Team and Enterprise, an Owner adds the connector first.

In Claude, go to Customize › Connectors, choose add, then Custom › Web.
Paste the server address and confirm:

https://api.app.membase.com/mcp-http

Your browser opens Membase's consent screen. Tick the Memory you made, and Profile access if Claude may read your profile. Select Approve access.
In a chat, open + › Connectors and switch Membase on.

Step 4: Connect ChatGPT

Your ChatGPT account or workspace must allow custom MCP servers.

Open ChatGPT Plugins (chatgpt.com/plugins), select + and choose Add custom MCP server.
Name it Membase, choose the public endpoint, paste the same server address, and set Authentication to OAuth. Leave Client ID and Secret empty.
Accept the warning, select Create as a plugin, and approve access on Membase's consent screen.
Install the plugin, start a new chat, type @ and pick Membase.

Step 5: Check that both can see it

In each assistant, ask "List my Membase containers." It should name your Memory. Then ask about something specific you added, such as "What did we decide about the launch date?", and check the answer.

If the list is empty, the Memory isn't switched on for that app: open Connect in Membase, select the app and switch the Memory on under Uses.

What ChatGPT and Claude can and can't do

They can read. Both can list the Memories you approved and search them.
They can't change your memory. Apps connected this way are read-only. Your chats reach Membase through the extension, not through the assistant.
They only see what you tick. Keep a separate Memory for personal notes and leave it off for work tools.
You can cut one off. Switch a Memory off for ChatGPT, or select Disconnect, and it stops on its next question; Claude keeps working.

Know that: Membase is in beta and free during the beta. The hosted service is not end-to-end encrypted. Learning happens when a Memory runs (on Update now or a schedule), not the instant you finish a chat.

Which option should you use?

Moving from ChatGPT to Claude for good: Option 1, Claude's import.
A few stable facts, no new tools: Option 2, a context file.
Using both every day, and wanting each to know what you did in the other: Option 3, a shared memory layer.
Also using Gemini, Grok, DeepSeek or Kimi: Option 3, through the browser extension.
Also using Claude Code, Cursor or Codex: Option 3. The same Memory connects to coding tools over MCP; see Membase Use Case 01.

Other memory tools that work across assistants include MemoryPlugin, AI Context Flow and Supermemory. For a comparison, see Best AI Agent Memory Tools in 2026.

Get started: Open Membase · Connect your AI · Browser extension · Discord

How to Share Memory Between ChatGPT and Claude (2026 Guide)