Do ChatGPT and Claude remember across chats? How to control it

Yes. ChatGPT has saved memories plus an option to reference your chat history. Claude generates memory from your chats, keeps it per project, and can search past chats on paid plans. Both let you view, edit and turn memory off in Settings. Neither shares memory with the other.

Does ChatGPT have memory across chats?

Yes, in two forms. According to OpenAI's help center, saved memories are a list of specific facts you asked ChatGPT to keep, and you can open that list and edit it. Reference chat history is different: it lets ChatGPT use relevant information from past conversations to personalize future answers, but it is not a list you can browse.

OpenAI also describes a newer Memory experience with a memory summary, and the controls you see depend on your account and plan. You can manage the pieces separately. OpenAI says that turning off Reference chat history does not delete your saved memories. It also says that information remembered from past chats is scheduled for deletion from OpenAI systems within 30 days when you turn it off, and that the original chats stay in your history until you delete them.

Does Claude have memory across chats?

Yes. Anthropic says memory is on by default for Free, Pro and Max plans on the web, Claude Desktop and Claude Mobile. On Team and Enterprise plans the owner controls whether it is available.

Claude's memory is organized by topic. In Settings, Memory, Topics you can read everything Claude remembers, then edit or delete any entry. Each project has its own separate memory and project summary, so one project does not leak into another or into your non-project chats.

Claude can also search your past chats on paid plans. You ask in plain words, such as what did we discuss about a topic, and it looks through earlier conversations. In projects, the search stays inside that project.

Does Claude share memory across chats, and with ChatGPT?

Within Claude, yes, unless you are in a project, where memory is kept separate for that project. Between assistants, no. ChatGPT's memory and Claude's memory are separate stores and neither reads the other.

Anthropic describes an experimental option to transfer memory between Claude and other AI services through Settings, Memory. Because it is marked experimental, check what it supports on your account before relying on it.

How to turn memory off

  1. In ChatGPT, open Settings and find the memory options (the section name can change). Switch off Reference chat history, saved memories, or both, depending on what you want to stop. Delete individual saved memories from Manage memories.
  2. In Claude, open Settings, Memory. Choose Pause memory to keep what exists but stop creating new entries, or Reset memory to delete all of it permanently.
  3. For a single Claude chat, click the plus button in the message box before your first message and switch memory off for that chat.
  4. For a fully private Claude chat, use an incognito chat from the ghost icon. Anthropic says those are not saved to memory or chat history.
  5. Menu names can change. If you do not see them, search the provider's help center for memory.

What to check before you rely on memory

Read the list. Saved memories can be wrong, out of date, or too personal for work chats. Delete what you would not want repeated.

Do not put secrets in memory. Passwords, health details and client data belong nowhere in a chat assistant's memory unless your plan and settings clearly allow it.

Memory is not a perfect record. It stores short facts, not your whole conversation. For serious projects, keep your own notes or a project file.

Where a node fits, honestly

A node does not replace ChatGPT's or Claude's built-in memory, and it cannot read it. Those stay with each provider.

A node gives you a memory of your own that every agent you connect can read. Agents that join share one thread and one memory, plus the files and notes you choose to put in. It runs on your own computer, and each step is written to one readable thread. That makes it useful when you switch between assistants and want the same facts everywhere.

Do it with a NodePress node

  1. Create a node on your computer and name it for the project or topic.
  2. Paste the one line into ChatGPT, Claude and any other agent you use.
  3. Tell the agents which facts to store in the node memory and which files to share.
  4. Review the thread and memory now and then, and remove anything that is wrong or private.

If you only use one assistant and its built-in memory already does what you need, you do not need a node. A node also does not import or read ChatGPT's or Claude's own memory.

Questions people ask

Does ChatGPT have memory across chats?

Yes. It has saved memories you can view and edit, and an optional setting to reference your past chat history. You can switch each off in settings.

Does Claude have memory across chats?

Yes. Memory is on by default on Free, Pro and Max. You can view it under Settings, Memory, pause it, or reset it. Paid plans can also search past chats.

Does Claude share memory across chats?

Yes outside of projects. Each project has its own separate memory and summary, so projects stay apart from each other and from regular chats.

How do I disable ChatGPT memory?

Turn off Reference chat history or saved memories in settings. OpenAI says turning off chat history does not delete saved memories, which you remove separately.

Can ChatGPT and Claude share the same memory?

No. They are separate. Anthropic describes an experimental memory transfer option in Claude, so check what your account supports.

Related

Sources

NodePress gives your AI agents one shared place on your own computer. Free to start, no sign-up. Make your first node.