ChatGPT Memory: What It Remembers and How to Control It

You told ChatGPT you're vegetarian three weeks ago, and today it quietly skipped the bacon in a recipe it wrote for you. That's ChatGPT memory at work. It can save details you share and pull context from older chats so you don't have to repeat yourself. It's useful, but it also raises fair questions about what gets stored, where it lives and how to wipe it. Here are straight answers to the questions people search most, based on OpenAI's current help pages.

What does ChatGPT memory actually remember?

ChatGPT memory remembers two kinds of things: details you've asked it to keep, and useful context it picks up from your past conversations.

OpenAI's Memory FAQ splits these into two settings:

• Reference saved memories covers facts you've explicitly shared, such as your name, your job or a diet preference. OpenAI says saved memories are always considered in future responses unless you delete them.
• Reference chat history lets ChatGPT look back at earlier conversations to recall things you mentioned along the way. These details aren't fixed, and OpenAI notes they can change over time.

You can trigger a saved memory on purpose. OpenAI's own example is telling ChatGPT, "Remember that I am vegetarian when you recommend a recipe." When something gets saved, you'll usually see a small "Memory updated" note in the chat.

OpenAI also says it tries to steer ChatGPT away from proactively saving sensitive details, such as health information, unless you ask it to. Sensitive information can still end up in memory if you share it, so think before you type.

Does ChatGPT memory work across chats?

Yes. That's the whole point of the feature. Saved memories aren't tied to a single conversation, so something you said in one chat can shape answers in a brand new one.

The second setting, Reference chat history, goes further. With it on, ChatGPT can draw on your previous conversations as a whole, not just the facts it formally saved. According to OpenAI, there's no storage limit on what it can reference this way.

How do I turn ChatGPT memory on or off?

Open Settings, then Personalization, and use the two memory toggles. You'll find Settings by tapping or clicking your profile picture.

The two switches are labeled Reference saved memories and Reference chat history. You can keep both on, use only saved memories, or turn both off. They're linked in one direction: switching off saved memories also switches off chat history.

What happens when you flip them:

1. Turning off saved memories stops ChatGPT from creating or using them, but it does not delete what's already stored.
2. Turning off chat history deletes the information ChatGPT remembered from past chats within 30 days, according to OpenAI.
3. Turning memory back on later picks up whatever saved memories you never deleted.

OpenAI's help pages describe the path as Settings, then Personalization, and also mention a Memory section in Settings, so the exact label can shift slightly between app versions. If you don't see Personalization, look for Memory instead.

How do I see and delete what ChatGPT remembers?

Go to Settings > Personalization > Manage memories to see every saved memory in a list, then delete the ones you don't want.

From that list, OpenAI's instructions are simple:

• To remove one memory, click the ⋯ menu next to it and choose Delete.
• To clear everything, click the ⋯ menu next to the search bar and choose Delete all.

You can also just ask. Typing "What do you remember about me?" gets you a summary, and telling ChatGPT to forget a specific detail works too. If you mostly use the app, our guide to using ChatGPT on iPhone covers where the main controls live.

Plus and Pro users on the web get a bit more. OpenAI says they can view a memory's version history from the three dot menu and restore an earlier version.

Can ChatGPT memory get full?

Saved memory can reach capacity, which is why some people see a "memory full" message. OpenAI doesn't publish an exact limit for saved memories.

To help avoid that, OpenAI added automatic memory management for Plus and Pro users on the web. It ranks memories by how recent they are and how often you bring up a topic. Memories that aren't top of mind appear in gray. You can turn automatic management off, and you can manually prioritize or deprioritize any single memory.

If you're on the free plan and hit the limit, the practical fix is cleanup. Open Manage memories, delete outdated or duplicate entries (old jobs, trips you already took, one time requests), and new memories can be saved again.

Reference chat history works differently. OpenAI says there's no storage limit on what ChatGPT can reference from past chats, so that side won't fill up.

Does deleting a chat erase the memory?

No. Deleting a conversation does not remove a saved memory that came from it.

OpenAI's help pages say that to fully remove something, you need to delete both the saved memory in Manage memories and the chat where you shared it. Doing only one leaves a copy behind.

Even after you delete a saved memory, OpenAI says it may keep a log of deleted saved memories for up to 30 days for safety and debugging. After that window, it's gone from their systems.

Is ChatGPT memory private, and is it used for training?

Memories stay tied to your account, but they can be used to improve OpenAI's models unless you opt out.

The Memory FAQ says that when Improve the model for everyone is turned on, your content, including saved memories, may be used for training. You can switch that off under Settings > Data Controls, which OpenAI explains in its Data Controls FAQ.

Business, Enterprise and Edu workspaces are not used for training by default, per OpenAI. In those workspaces, the owner can turn memory on or off for everyone, and OpenAI says Reference chat history isn't yet available to Enterprise and Edu customers.

If you share a device or worry about someone reading your chats, memory isn't the only setting to check. Our walkthrough on ChatGPT locked chats covers another layer of protection.

Is memory different on the free plan?

Yes, a little. Free users get a lighter version of memory, while Plus and Pro get deeper, longer term memory and more controls.

OpenAI described the free version as giving short term continuity across conversations, while Plus and Pro get a longer term understanding of the user. The Memory FAQ also says memory management controls and automatic memory management are currently only on the web for Plus and Pro.

The sources memory can draw from also vary by plan, according to OpenAI:

• Free and Go: past chats, saved memories and custom instructions.
• Plus and Pro: all of the above, plus files and Gmail (Gmail isn't offered in the EEA, Switzerland or the UK).

If memory is a big reason you're weighing an upgrade, our ChatGPT Plus vs free comparison looks at the other differences that matter more day to day.

How do I use ChatGPT without memory for one conversation?

Start a Temporary Chat. It doesn't read your existing memories and doesn't create new ones.

According to OpenAI's Temporary Chat FAQ, you start one by tapping the model name at the top of the screen and choosing Temporary chat. These chats don't show up in your history and aren't used to train models, though OpenAI says it may keep a copy for up to 30 days for safety. Your custom instructions still apply if you have them turned on.

It's the easier option for one off questions you'd rather not have shape future answers, since you don't have to switch memory off and remember to turn it back on.

Should you keep ChatGPT memory on?

For most people, yes, with a quick checkup now and then. Memory makes ChatGPT less repetitive and better at things like recipes, writing in your voice or keeping track of a project.

A good habit: once a month, open Manage memories and skim the list. Delete anything outdated or anything you'd be uncomfortable seeing in a data export. If you're new to the tool, start with our plain English explainer on what ChatGPT is, or browse more guides in our AI section.