ChatGPT Memory feature is now live: chat preferences can be saved and managed with one click

ChatGPT’s Memory feature cuts out the hassle of “having to repeat yourself every time”: it can reference the preferences and background you’ve mentioned in later conversations, making its answers better match your habits. More importantly, ChatGPT Memory isn’t forced on—you can view, delete, or turn it off at any time.

What exactly does ChatGPT Memory remember?

The core of ChatGPT Memory is to retain “stable information that’s useful to you,” such as your writing tone, commonly used formats, and the background of long-term projects you’re working on. You can also say explicitly in a conversation, “Please remember that I like summaries in tables,” so ChatGPT Memory can save preferences like that.

Note that memory is not the same as the chat history itself: deleting a specific chat doesn’t necessarily remove the saved memories along with it. To completely remove a particular item, you need to delete it separately on the ChatGPT Memory management page.

How to view and manage ChatGPT Memory

When ChatGPT Memory updates, the interface usually shows a prompt like “Memory updated,” and you can click straight into the management entry to see what was added. After you go into Settings and find the personalization-related options, you’ll see a list for “Manage memories.”

On the management page, it’s a good idea to clean up regularly: delete outdated work information and finished projects to prevent ChatGPT Memory from citing old context in new tasks. You can also simply ask in a chat, “What do you remember about me?” as the easiest way to do a quick review.

Two ways to turn it off: turn off memory or turn off chat referencing

If you want ChatGPT to be entirely “this conversation only, ends when it ends,” you can turn off “Saved memories” in ChatGPT Memory. In addition, some accounts will also see a “Chat history” toggle, which controls whether it can reference past conversations to produce personalized responses.

These two toggles mean different things: the former is more like “archiving long-term preferences,” while the latter is more like “deriving habits from conversation history.” If you’re not sure which to choose, you can turn off chat referencing first, while keeping only a small number of memories you explicitly asked it to save.

Temporary chats: keep ChatGPT Memory out of this conversation

When dealing with sensitive topics or one-off needs, temporary chat mode is recommended: it won’t appear in your history and it won’t update ChatGPT Memory. This lets you ask normally while keeping information boundaries clear.

For everyday use, a simple combination is: use ChatGPT Memory for long-term projects to improve continuity, and use temporary chats for quick questions to preserve privacy and keep the context clean. Once you build this habit, ChatGPT becomes smoother to use—and more controllable.