Importing Your ChatGPT History into Claude is the migration workflow that moves a user's 📝ChatGPT conversation archive into a 📝Claude project, giving Claude reasoning access to prior sessions.
The technique works on consumer ChatGPT plans (Free, Plus, Pro) where users can self-serve a data export from OpenAI. ChatGPT Team and Enterprise seats cannot export without admin approval, leaving many paid users locked out of their own conversation history. Once the export is triggered, OpenAI emails a time-limited download link with a ZIP archive containing conversations, custom instructions, and memory state.
On the 📝Anthropic side, the unzipped files upload to a Claude Project's knowledge base, where they become retrievable context during every chat in that project. The user can then query Claude with full awareness of prior ChatGPT threads — past tool use, prompts, decisions, and stylistic patterns — without re-training or re-explaining anything.
The technique surfaced publicly via Taryn Southern's January 2026 viral reel and has since become the canonical bridge for users migrating away from ChatGPT. It works because Claude's project model rewards portable, structured memory in a way ChatGPT's per-thread architecture does not.
The migration comes up on nearly every onboarding call I run — Plus users export cleanly, Team and Enterprise users get locked out of their own data, and that lockout has become a forcing function for several people in my orbit to leave ChatGPT entirely.
Tutorial
- In ChatGPT, go to Settings → Data Controls → Export Data.
- Confirm via the OpenAI email — your archive link expires in 24 hours.
- Download and unzip the archive; locate
conversations.jsonandchat.html. - In Claude, create a new Project or open an existing one.
- Open the Project's knowledge base and upload the extracted files.
- Chat with the Project — Claude now references your prior ChatGPT history.
- Optionally connect 📝MythOS via 📝MCP to layer structured memory on top.