Guide · Updated August 2026
Claude import memory — and everything memory missed
Claude's official import memory feature pulls your ChatGPT memory over for free in about a minute: open Settings → Capabilities in Claude, start the import, copy the prompt Claude gives you into ChatGPT, and paste ChatGPT's packaged reply back. Use it — it's the fastest legitimate transfer there is. But understand what it moves: the few dozen facts ChatGPT chose to remember about you, not the months of context sitting in your conversations. Your projects, decisions, working styles and the hundred small corrections you made live in conversation history, and memory never saw most of them. So treat the import as step one, not the whole move: bring the memory layer over in a minute, then transfer the history layer with a summary prompt or a distilled context document. This page covers both layers — how the official import works, what it misses, and how to move the rest.
How Claude's import memory feature works
- In Claude, open Settings → Capabilities and find Import memory from other AI providers → Start import.
- Claude shows a prompt. Copy it.
- Paste the prompt into ChatGPT. It replies with a packaged summary of everything its memory has stored about you.
- Copy that reply, paste it back into Claude's import window, and click Add to memory.
Free on all plans since March 2026. Also works from Gemini and Copilot. Anthropic labels the feature experimental — imported personal details unrelated to work may not always stick.
What ChatGPT's memory actually contains
Memory is a short list of facts ChatGPT decided to save: your name, role, a few preferences, some recent projects. It's recency-biased and selective by design. Everything else — the project you spent March on, the decision you reasoned through in twenty messages, the people you mention weekly, how you actually write — lives only in your conversation history. For most long-time users, memory holds a page; the history holds a book.
Does Claude have memory like ChatGPT?
Yes. Claude's memory (Settings → Capabilities) works across chats like ChatGPT's does, is available on free accounts, and is fully viewable and editable — you can read every stored entry, correct it, or delete it. The practical differences: Claude's memory skews toward work context by design, projects keep their own scoped memory, and — unlike ChatGPT — Claude can import memory from other assistants. What neither memory holds is your conversation archive, which is where the next section comes in.
Import vs summary prompt vs full history
| Memory import | Summary prompt | Full history | |
|---|---|---|---|
| Source | Stored memory facts | One recall pass | Every conversation |
| Old projects | Mostly missing | Recency-biased | Captured |
| People & decisions | Rarely stored | Partial | Extracted |
| Writing style | No | Vague | Analyzed |
| Effort | 1 minute | 15 minutes | Upload a ZIP |
| Cost | Free | Free | Free for 3 convos · from €9.99 |
These aren't competing options — do the one-minute import for the basics, then transfer the full history for everything it missed.
Transfer everything memory missed
- Export your ChatGPT data (Settings → Data controls → Export data).
- Upload the ZIP to ChatGPTToClaude — no account needed.
- Get one context document distilled from your entire history and paste it into Claude's profile preferences or a Project.
Frequently asked questions
Is Claude's memory import free?+
Yes. Since March 2026 it's available to all Claude users, including free accounts, under Settings → Capabilities → Import memory.
Does the memory import move my conversations?+
No. It transfers only the summary of facts ChatGPT's memory stored about you. Conversations, Projects, Custom Instructions and Custom GPTs are not included.
Can I import memory from Gemini or Copilot too?+
Yes — Claude's import flow supports ChatGPT, Gemini and Copilot with the same paste-a-prompt mechanism.
Does memory stay in sync after importing?+
No. The import is one-time. New memories ChatGPT creates afterwards don't appear in Claude — you'd need to re-run the import (or re-run a full-history transfer for a complete refresh).
Does Claude carry memory across chats?+
Yes. With memory enabled (Settings → Capabilities), Claude remembers context across conversations, and you can view and edit what it has stored. Project conversations keep their own project-scoped context.
Does Claude have permanent memory?+
Memory persists until you edit or delete it — it doesn't expire on its own. It's also under your control: you can review every entry, correct it, or wipe it entirely in settings.
Is it good to turn on memory in Claude?+
For most people, yes — it's what makes an imported context stick and saves you re-explaining yourself. Turn it off if you share the account or regularly discuss things you don't want carried forward; you can also delete individual memories instead.
The complete walkthrough: How to switch from ChatGPT to Claude →