Claude memory vs ChatGPT memory: which one lets you take it with you

Claude and ChatGPT both remember you, but neither lets the other assistant use what it remembers. Most comparisons score them on features. We think the better question is exit, so here is our own proposal, the exit test: for any AI memory, ask what you can take out, in what format, and whether another assistant can use it without you retyping it.

What each one remembers

ChatGPT keeps two things, according to OpenAI's Memory FAQ: saved memories, which are explicit details stored apart from your chats so deleting a chat does not remove them, and a reference to your chat history that can change over time. You manage them under Settings, Personalization, Memory.

Claude's memory, as Anthropic described it on August 25, 2026, is stored as short files under Topics in Settings, Memory. You can read, edit or delete each one, and it is on by default on Free, Pro and Max across web, desktop and mobile. Claude has other memory stores too, which we compare in Claude memory explained.

The exit test, scored

Why moving by copy and paste loses something

Anthropic's import steps work like this. Claude gives you a prompt to run in your old assistant, which lists your memories in one block of text. You paste that block into Settings, Memory, and Claude extracts key details and stores them as separate entries. Anthropic says Claude's memory focuses on work topics, so unrelated personal details may not be kept, and that not every memory may be incorporated.

Our reading, not Anthropic's claim, is that this makes two summarizing passes. The first assistant writes a summary of what it remembers, and the second extracts from that summary. What arrives is a summary of a summary, filtered for work topics, and you cannot compare it against the original in one place because the original stays inside the other app.

What a memory you own looks like

A memory you own is stored on your computer, readable by more than one assistant, and exportable as plain files. ModelBrain is our attempt at that. The export command writes every memory out as Markdown or JSON Lines, and only the current version of each one: a memory you replaced is named in the supersedes field of the memory that replaced it. The free tier includes full export at any time.

You can use the same vault in Claude and ChatGPT with no copy step, because Claude Desktop, Claude Code and Cursor talk to it directly and ChatGPT reaches it through a small relay we run. How that route works, and what it can and cannot do on each ChatGPT plan, is in Can ChatGPT use a memory that lives on your computer?

Where ModelBrain costs you something

ModelBrain is in beta and its engine is not open source today. The free local vault is not encrypted at rest, and exported files are plain unencrypted text, so keep them somewhere you trust. ChatGPT on the web needs a free ModelBrain account for the relay. Syncing the same memory across computers is paid, at $6 a device a month billed yearly. If you only ever use one assistant, that assistant's own memory has less to set up. For what each part of this stores and where, see privacy.

Which to choose

Use ChatGPT's memory if you want automatic recall across chats and you stay inside ChatGPT. Use Claude's if you want visible topic files and the scoping Projects give you. Use a memory you own if you work in more than one assistant, or if you want the option to leave without retyping everything.

Sources

• Memory FAQ, OpenAI Help Center
• Import and export your memory from Claude, Anthropic support
• Claude's memory works everywhere, and you decide what's in it, Anthropic, August 25, 2026
• Export and search your memories from the command line, ModelBrain docs
• Pricing, ModelBrain

Common questions

Can I export my ChatGPT memory? OpenAI's Memory FAQ, checked October 22, 2026, describes viewing, editing and deleting memories under Settings, Personalization, Memory, and does not describe an export of saved memories. You can ask ChatGPT what it remembers and copy the answer.

Can I move my ChatGPT memory to Claude? Yes, as text. Claude's import gives you a prompt to run in ChatGPT, then you paste the result into Settings, Memory. Anthropic calls imports experimental and says not every memory may be incorporated.

Which is better, Claude memory or ChatGPT memory? It depends on what you value. ChatGPT's recalls across all chats automatically, and Claude's is organized as editable topic files. Neither can be read by the other assistant, so if you use both, a memory you own avoids the copy step.

Last checked October 22, 2026. Settings and tool behavior change often, so confirm a detail on the vendor's own page before relying on it.