Does Claude Have Memory? How It Works in 2026

Yes. Claude has a memory: it saves topics from your chats and uses them in later conversations. Unlike ChatGPT and Gemini, it shows you a list of what it remembers, lets you edit or delete it one topic at a time, and documents a way to export it. Everything below comes from Anthropic's own help pages and release notes, read on 4 October 2026.

The short answer

• Does Claude remember past chats? Yes, when memory is on. Claude saves individual topics while you chat, and you can also ask it to remember something.
• Is it on by default? On Free, Pro and Max (web, desktop and mobile), yes. On Team and Enterprise, an owner decides whether memory is available, and it stays off for each member until they turn it on.
• Can you see what it remembers? Yes. Settings › Memory lists every topic.
• Can you take it with you? Yes, with an asterisk: the export is prompt-based and Anthropic labels import and export experimental.

How Claude's memory works

Claude builds memory as topics: individual things it has learned about you or your work, saved while you chat. You can also tell it directly to remember something.

A few rules from Anthropic's help page:

• Projects are separate. Each project has its own memory space, so what Claude learns in one project stays there.
• Sensitive topics are excluded unless you opt in. Some data is never saved at all, including government IDs, criminal history, financial account numbers and immigration status.
• Search past chats (Claude looking through your earlier conversations on request) is limited to paid plans.
• Claude and Cowork share memory in the cloud since 25 August 2026. Local Cowork sessions don't use memory.

Source: Anthropic: Claude memory.

How to see, edit and delete Claude's memory

• See it: Settings › Memory lists every topic Claude has saved.
• Edit or delete: one topic at a time, from the same list.
• Pause: keeps your memory but stops Claude using it.
• Reset: deletes everything Claude has remembered.

One detail that surprises people: deleting a conversation does not remove the memory topics that came from it. To forget something, delete the topic itself.

Chats Claude won't remember

• Incognito chats are never remembered and don't stay in your history.
• The per-chat Memory switch in the "+" menu turns memory off for one conversation. Unlike incognito, that chat still stays in your history.

Moving your memory in and out

Claude is the only one of the three big assistants that documents a way out:

• Export: you ask Claude to write out its memories verbatim and copy the result. Your memory is also included in Claude's data export.
• Import: you paste memories from another assistant (for example, ChatGPT) into Claude's import box. Imports are available on Free, Pro, Max and Team, on web and desktop. Anthropic notes that details unrelated to work may not be kept.

Both are labelled experimental. Source: Anthropic: import and export memory.

Claude Code has its own, separate memory

If you use Claude Code, it doesn't use the memory described above. It has two systems of its own, both loaded at the start of every session:

• CLAUDE.md files that you write (per user, per project, or local to one checkout).
• Auto memory that Claude writes for itself, under ~/.claude/projects/<project>/memory/.

Auto memory is machine-local: Anthropic's docs say it isn't shared across machines or cloud environments. Type /memory inside Claude Code to see which memory files are loaded, open the auto memory folder, or turn auto memory off. Source: Claude Code docs: memory. For tools that extend it, see memory tools for Claude Code.

How Claude's memory compares with ChatGPT and Gemini

The full side-by-side is in which AI has the best memory. The same answer for the other two: how ChatGPT memory works and does Gemini have memory like ChatGPT.

If you use more than one assistant

Claude's memory, like every built-in memory, is readable only inside Claude. Cursor, ChatGPT and Gemini can't see it, and Claude Code doesn't share it either.

If what you want remembered is your own knowledge (the videos, articles and documents you rely on), the alternative is to keep it in one store outside every assistant and connect each of them to it over MCP. BrainTube is ours, so read this as labelled: it keeps what you save in one library that Claude, Claude Code, Cursor and other MCP clients can search and save to. It doesn't replace the profile-style memory described above. Setup is in how to give Claude long-term memory with MCP, and other options are in Claude memory alternatives.

FAQ

Does Claude remember previous conversations?

Yes, when memory is on. It saves topics from your chats and uses them in later conversations. Paid plans can also search past chats on request.

Does Claude have memory on the free plan?

Yes. Memory is on by default on Free, Pro and Max. On Team and Enterprise, an owner decides whether it's available, and each member then turns it on.

How do I make Claude forget something?

Open Settings › Memory and delete the topic. Deleting the conversation it came from is not enough. To wipe everything, use Reset.

How do I stop Claude remembering a chat?

Start an incognito chat, or turn memory off for that chat with the Memory switch in the "+" menu.

Can I move my ChatGPT memory to Claude?

Yes. Ask ChatGPT to write out what it remembers, then paste it into Claude's memory import. Anthropic labels this experimental; OpenAI documents no export of its own.

Does Claude Code remember between sessions?

Yes, through CLAUDE.md files and auto memory, both loaded at the start of each session. Auto memory stays on the machine where it was written.