Last checked: August 26, 2026
If you have spent months teaching ChatGPT how you work, what you prefer, and which instructions matter, trying another AI assistant can feel like starting from zero. Fortunately, you can now transfer ChatGPT memory to Claude using Claude’s built-in memory import workflow.
The important part is understanding what “transfer” means. Claude does not clone your ChatGPT account. Your full chat sidebar, uploaded files, ChatGPT Projects, custom GPTs, subscription, and account settings do not automatically appear inside Claude.
Quick answer: You can transfer ChatGPT memory to Claude by generating a useful context summary in ChatGPT, reviewing that information, opening Claude’s memory import tool, and adding the cleaned context to Claude. Claude then extracts useful information into its own memory entries.
This guide explains the complete process, what actually moves, what stays behind, how to avoid importing unnecessary private information, and how to check what Claude really learned after the transfer.
Can You Transfer ChatGPT Memory to Claude?
Yes. Claude now has a built-in import flow designed to bring memory and useful context over from other AI providers.
For users on Claude’s newer memory experience, the route is Settings → Memory → Start import. Accounts still using the legacy memory interface may instead see Settings → Capabilities → Memory.
When you transfer ChatGPT memory to Claude, the process is closer to a reviewed context handoff than a traditional data migration:
memory & context
the information
its own memories
Claude is not signing into your ChatGPT account behind the scenes. Instead, you generate a memory/context summary in your current AI service, inspect it yourself, and then submit that information to Claude.
Memory transfer is not account cloning. The goal is to carry useful context forward so Claude does not have to learn everything from scratch.
What Happens When You Transfer ChatGPT Memory to Claude?
Claude processes the context you provide and extracts useful information into individual memory entries. That means the destination memory may not be an exact word-for-word copy of the source.
Anthropic also says memory import remains experimental. Some submitted information may not be incorporated successfully, and Claude’s memory is designed to focus particularly on context that helps it collaborate effectively on work-related topics.
| ChatGPT information | Expected result | What it means |
|---|---|---|
| Writing and response preferences | Good candidate | Formatting, tone, answer length, and other stable preferences can be useful to carry over. |
| Recurring instructions | Good candidate | Repeated rules about how you like AI to help can provide useful context. |
| Ongoing goals | Potentially | Claude may retain ongoing goals when they are useful for future work. |
| Tools and workflows | Potentially | Software, frameworks, and recurring processes can be useful working context. |
| Every ChatGPT conversation | No | Your old conversations are not recreated as Claude conversations. |
| ChatGPT Projects | Not recreated | Project chats, structure, files, and workspace organization remain separate. |
| Uploaded files | Not automatic | Files you still need must be moved separately. |
| Custom GPTs | No | A custom GPT is a separate configured assistant and is not reproduced by memory import. |
| ChatGPT settings | No | Account-level settings remain with OpenAI. |
| Subscription or billing | No | Claude and ChatGPT accounts remain completely separate. |
So when you transfer ChatGPT memory to Claude, useful context can move, but the products, files, conversations, and account structures surrounding that context generally do not.
Before You Transfer ChatGPT Memory to Claude
Do not copy the first memory summary ChatGPT produces and immediately paste it somewhere else.
Before you transfer ChatGPT memory to Claude, treat the exported context as information you are deliberately sharing with another provider. Review every item and ask whether Claude genuinely needs it.
- Remove outdated information and completed projects.
- Correct memories that are inaccurate or contradictory.
- Remove passwords, login details, security codes, and financial information.
- Remove unnecessary confidential or client-specific material.
- Keep stable response preferences that still matter.
- Keep recurring work instructions that save you time.
- Keep relevant goals, workflows, tools, and project context.
This is also a useful opportunity to clean up ChatGPT itself. If the source memory contains information you no longer want there, use our guide on how to delete ChatGPT memory completely before or after the migration.
Privacy reminder: Moving context between AI services means intentionally providing information to another provider. Transfer only what you are comfortable sharing and actually need for future use.
How to Transfer ChatGPT Memory to Claude Step by Step
The safest way to transfer ChatGPT memory to Claude is to use Claude’s official import workflow while keeping yourself in control of the information that crosses between services.
Open Claude on the web or Claude Desktop and go to Settings → Memory. If you are using Claude’s newer memory experience, select Start import.
If your account still shows memory inside Settings → Capabilities, you are on the legacy experience. That interface also provides an import option.
Claude’s import flow provides a prompt designed to help another AI service list stored memories and useful learned context. Copy the current prompt from Claude and run it in ChatGPT.
If you prefer a more selective starting point, you can also use the privacy-conscious prompt below and then compare the result with the information Claude’s official workflow asks you to collect.
I am preparing a context summary to use with another AI assistant. List the information you currently remember or consistently use about me that would be useful for future work. Include: - how I prefer answers to be written or formatted - recurring instructions I have given you - ongoing professional goals or projects - tools, software, frameworks, or workflows I regularly use - useful preferences and corrections I have made - recurring topics that materially affect how you should help me Do not invent missing information. Separate each item clearly and flag anything you are uncertain about. Avoid including passwords, authentication information, financial data, or sensitive information that is unnecessary for future work.
Read the result carefully. Remove information that is obsolete, wrong, private, repetitive, or irrelevant.
This is one of the biggest advantages of a manual context handoff: you can improve the information before Claude ever sees it.
Return to Claude’s import screen, paste the cleaned context, and select Add to memory.
This is the point where you actually transfer ChatGPT memory to Claude. Claude processes the submitted context and extracts information into its own memory entries.
After the import completes, review Claude’s memory rather than assuming everything transferred correctly.
Claude’s newer experience includes a way to start a conversation about what it learned from the updated memory. You can also ask directly.
Review what you currently remember about me after the import. Organize it into: 1. response preferences 2. recurring instructions 3. work and projects 4. tools and workflows 5. ongoing goals 6. other useful context Flag anything that appears incomplete, contradictory, vague, or outdated.
If an important detail is missing, add or correct that specific information rather than importing the entire source again.
In Claude’s newer memory experience, you can use its memory controls to tell Claude what to change or remove. You can also make corrections conversationally.
Best practice: Use migration as a cleanup opportunity. A smaller collection of accurate, useful context can be more valuable than importing years of stale instructions.
ChatGPT Memory Is Not the Same as Chat History
This distinction matters because people often use “memory” and “history” as though they mean the same thing.
OpenAI’s current memory system can synthesize useful context from past chats, while individual conversations still exist separately in chat history. In the legacy saved-memory system, saved memories are also stored separately from the chats where they originated.
That is why you can transfer ChatGPT memory to Claude without transferring your complete conversation archive.
- Your old ChatGPT conversations remain in ChatGPT unless you separately delete them.
- Claude does not recreate those conversations in its sidebar.
- A remembered preference can move as context even though the original chat does not.
- Deleting a chat does not always mean associated saved memory disappears automatically.
If you are actually trying to find an old conversation, uploaded document, or previous answer, see our guide on how to search ChatGPT history and find old chats and files.
Does a Full ChatGPT Data Export Transfer Everything to Claude?
No. A ChatGPT data export and Claude’s memory import solve different problems.
OpenAI lets eligible signed-in users request account data through ChatGPT’s settings or the Privacy Portal. A downloaded export can include chat history and other relevant account data.
That makes the export useful for backup, records, and certain manual migration tasks—but Claude’s memory import does not automatically turn that archive into a fully recreated Claude account.
| Method | Best use | Creates a full Claude copy? |
|---|---|---|
| Claude memory import | Move useful preferences, instructions, goals, and context | No |
| ChatGPT data export | Keep an archive of eligible account data and chat history | No |
| Manual file migration | Move documents needed for future work | Partial |
| Manual workflow reconstruction | Rebuild Projects, instructions, or specialized setups | Partial |
Can ChatGPT Projects Transfer to Claude?
Not as complete working Projects.
A ChatGPT Project can include its own instructions, conversations, files, and project-specific context. Memory import does not recreate that entire structure inside Claude.
Even after you transfer ChatGPT memory to Claude, important ChatGPT Projects should be reviewed and rebuilt selectively if you still need them.
- Write down the project’s permanent instructions.
- Identify the source files still required.
- Summarize the current project state.
- List unresolved tasks or decisions.
- Move only the files you still need.
- Recreate relevant instructions in the destination workspace.
- Confirm that Claude understands the project before continuing important work.
For more context on how ChatGPT structures differ, see ChatGPT Projects vs Custom GPTs.
Can Custom GPTs Transfer to Claude?
No. Custom GPTs are separate configured assistants rather than simple memories.
A custom GPT can contain instructions, uploaded knowledge, capabilities, and other settings. You can transfer ChatGPT memory to Claude, but Claude’s memory-import tool does not clone a custom GPT or reproduce its complete configuration.
If a custom GPT contains a workflow you depend on, separate the parts:
Why Some Memory May Be Missing After the Transfer
If you transfer ChatGPT memory to Claude and some information is missing afterward, that does not necessarily mean you performed the steps incorrectly.
1. Memory import is still experimental
Anthropic says memory imports remain in active development and may not always incorporate everything submitted.
2. Claude prioritizes useful working context
Claude’s memory is designed to emphasize information that helps it collaborate effectively. Personal details that are unrelated to work may not be retained in the same way.
3. The ChatGPT summary may be incomplete
OpenAI explains that ChatGPT’s memory can represent a broader synthesis of past context than what appears in an individual memory summary. If the source output omits something, Claude cannot import information it never received.
4. Your information may conflict
Long-term AI use often creates contradictory instructions. For example, an old instruction might request extremely short answers while a newer one asks for comprehensive explanations.
5. Temporary information may not deserve long-term memory
Not everything useful in one conversation needs to become permanent cross-platform context.
Who Benefits Most From Transferring AI Memory?
The strongest use cases involve people who have built up reusable working context rather than those who only ask occasional one-off questions.
Freelancers and consultants
Useful context may include writing preferences, recurring deliverable formats, preferred tools, workflow rules, and broad project conventions. Do not casually move confidential client information.
Small-business owners and marketers
You might carry over recurring planning formats, brand-neutral working preferences, content processes, marketing workflows, or the way you evaluate campaigns and ideas.
Developers and technical professionals
Preferred languages, frameworks, coding conventions, explanation style, and recurring environment constraints can reduce setup time when trying a new assistant.
Students and researchers
Study preferences, explanation level, citation expectations, and recurring research themes may be useful. Actual papers, readings, datasets, and uploaded files need separate handling.
Designers and creators
Useful context can include preferred output structure, review criteria, production constraints, workflow steps, or recurring creative requirements—provided those instructions do not include material you should not share elsewhere.
Should You Transfer Everything ChatGPT Knows?
Usually not.
Deciding whether to transfer ChatGPT memory to Claude should begin with relevance, not completeness.
| Context type | Recommendation | Why |
|---|---|---|
| Stable response preferences | Usually keep | They can improve future answers immediately. |
| Current working style | Usually keep | Helpful for repeated workflows. |
| Ongoing goals | Keep if current | They can reduce repeated explanations. |
| Completed projects | Usually remove | Old context can clutter future responses. |
| Outdated preferences | Remove | They can actively make answers worse. |
| Contradictory memories | Fix first | Conflicting instructions reduce consistency. |
| Passwords or authentication data | Do not transfer | They are not useful AI memory. |
| Unnecessary sensitive details | Avoid | Only move information required for your intended use. |
A Better Way to Transfer ChatGPT Memory to Claude
The goal when you transfer ChatGPT memory to Claude should be to move context that improves future answers—not to preserve every detail ChatGPT has ever learned.
Use this six-part migration framework:
- Extract: generate the useful context ChatGPT currently uses.
- Inspect: read every item before sharing it elsewhere.
- Clean: remove obsolete, incorrect, unnecessary, or sensitive information.
- Import: add the reviewed context through Claude’s official memory tool.
- Verify: ask Claude what it retained.
- Rebuild selectively: move Projects, files, or specialized workflows only when they are still valuable.
The useful mental model: migrate your operating context, not your digital clutter.
If you frequently notice AI assistants losing earlier context even without switching services, read our guide on why AI forgets earlier messages and how to stop losing context.
Frequently Asked Questions
Can I transfer ChatGPT memory to Claude?
Yes. Claude provides a built-in memory import workflow that lets you bring useful context from another AI provider into Claude. The process involves generating the source context, reviewing it, and submitting it through Claude’s import tool.
Does transferring ChatGPT memory delete anything from ChatGPT?
No. Importing information into Claude does not automatically delete your existing ChatGPT memory, chats, files, or account data.
Can Claude import my complete ChatGPT history?
Not through the memory-import feature. Memory transfer and conversation-history migration are different processes.
Can Claude import ChatGPT Projects?
Not as fully recreated Projects. Relevant project context can potentially be carried over as useful memory, but files, chats, structure, and project configuration need separate handling.
Can I transfer a custom GPT to Claude?
No automatic conversion exists through memory import. You would need to rebuild appropriate instructions, files, and workflows separately.
Why don’t I see Start import in Claude?
Claude is transitioning between memory experiences. Newer accounts may show Settings → Memory, while legacy users may still find Memory under Settings → Capabilities.
Is Claude memory import available for free users?
Anthropic currently says memory imports are available for Free, Pro, Max, and Team plans on Claude web and Claude Desktop.
Will Claude remember everything I import?
No guarantee. Anthropic describes memory import as experimental and says some imported information may not be incorporated successfully.
Should I remove private information before importing?
Yes. Review the source output before providing it to another AI service and remove anything unnecessary for your intended use.
Final Verdict
You can now transfer ChatGPT memory to Claude without manually reteaching every useful preference, instruction, goal, and recurring workflow from scratch.
But the strongest way to use the feature is to stop thinking of it as a full account migration.
Your conversations, Projects, files, custom GPTs, settings, and subscription remain separate. What you are moving is useful context.
When you transfer ChatGPT memory to Claude, review that context first, remove anything unnecessary, import the cleaned version, and then verify what Claude retained. That gives you a much cleaner starting point than blindly copying years of accumulated AI history.
