Does ChatGPT remember past conversations?

Short answer: yes — more than it used to. ChatGPT saves facts about you and draws on your earlier chats. The catch is the boundary: it remembers what passed through ChatGPT, and reaches further only if you connect a whole account to it. Hand it the thread instead, and the five emails before your meeting are simply there.

Yes. ChatGPT remembers past conversations in two ways: the facts it stores about you (a saved-memory list or, on the current system, a running memory summary) and the ability to draw on your earlier chats when memory is switched on. Both only cover what has passed through ChatGPT itself. It can't recall an email, document, or note that never went into it.

The honest answer: yes, with two mechanisms

A lot of older write-ups still say "ChatGPT forgets everything between chats." That hasn't been true for a while. Today ChatGPT remembers past conversations in two distinct ways, and it's worth knowing which is which.

• Saved memories. ChatGPT notices and stores specific facts about you over time — your name, what you do, projects you've mentioned, how you like answers formatted. You can open the memory settings and see them listed, edit them, or delete them one by one.
• Context from your past chats. ChatGPT can also draw on your past conversations as a whole, not just the facts it explicitly saved. OpenAI describes the current system as "a continually updated synthesis of context from your past chats," so a new chat can pick up relevant context from things you discussed weeks ago without you repeating yourself.

Turn both on and ChatGPT genuinely feels like it knows you. Ask it to draft something "in my usual style" and it often can. That's real, and it's useful.

Where the memory stops

Here's the part the headline answer skips. Everything ChatGPT remembers, it remembers from one place: your ChatGPT chats and what you've uploaded into them. Unless you've connected your Gmail or Drive to it account-wide, everything it knows comes from inside ChatGPT — and even then it's a full-account handover, not the five threads that matter. That boundary is the whole story.

If you're a consultant juggling seven clients, or anyone whose real work lives in email threads, shared docs, and meeting notes, the gap shows up fast:

• It doesn't know what was in the five emails that landed before your 9 a.m. call — not unless you pasted them into a chat, or handed it the whole inbox.
• It doesn't remember the PDF a client sent Tuesday, or the contract redline buried three replies deep in a thread.
• It doesn't have the voice memo you dictated walking to the car, or the screenshot you saved of a slide.
• It can't tell you what the CFO actually said about pricing in the October review, because that's in a transcript, not a ChatGPT conversation.

ChatGPT remembers what you told ChatGPT. It doesn't remember your life. To reach further it needs a whole account connected — your inbox, your drive — when what you wanted was to hand it one thread.

Three more limits worth knowing

Beyond the source boundary, there are three practical edges people run into:

• It's opaque. Saved memories you can inspect, but when ChatGPT references your broader chat history, there's no complete trail — the per-answer sources view shows some of what it pulled in, but by OpenAI's own description not all of it, and never the original source outside ChatGPT.
• Removing it is a sweep, not a switch. By OpenAI's own account, fully deleting something ChatGPT may know about you means clearing it from every source it appears in — past chats, archived chats, files, the memory summary, and any connected apps that hold it. Deleting a chat doesn't remove what was distilled from it, and deleting a memory doesn't scrub the chats. Here's what deleting a chat actually removes.
• It's per-account, per-app. The memory lives in your ChatGPT account only. It doesn't follow you to Claude, to your notes app, or anywhere your context actually scatters across the day.

How to see what ChatGPT remembers about you

There are two routes, and it's worth using both.

• Open the memory settings. Go to Settings → Personalization → Memory. What you find depends on which memory system your account is on. The newer system shows a memory summary — a running, plain-language write-up of what ChatGPT has picked up about you, stamped with when it was last updated. You edit it by typing a correction into the box beneath it or highlighting a line you want changed. Accounts on the older saved memories list (OpenAI lets you switch back to it from the same screen) see each stored fact as its own entry, deletable one at a time or all at once.
• Ask it directly. Type "What do you remember about me?" into a normal chat. OpenAI's own guidance is that the summary won't include everything ChatGPT remembers, and that asking in chat is the way to check whether a specific detail was kept.

There's also a per-answer view: the small book icon beneath a response lists what personalized it — custom instructions, past chats, files, memories — and tapping a memory explains why it was used. Useful, but partial by OpenAI's own description: it "may not show every factor or source that shaped a response." So the summary and the sources view between them show you a lot, and neither is the complete list. What ChatGPT draws from your wider chat history has no exhaustive inventory you can open. That's the opacity limit described above.

How to make ChatGPT forget previous conversations

There are four levers, from narrowest to widest. Use the smallest one that solves your problem — the wide ones take continuity away everywhere, including where it was helping.

• Correct or remove one detail. On the current system you edit the memory summary in place: OpenAI's documented routes are typing the change "into the text box at the bottom of the memory summary," or highlighting "any text in the memory summary to make a specific correction." On the legacy saved-memories list you delete the entry itself, and OpenAI says ChatGPT will also update, combine or remove saved memories when asked. One catch either way: removing a memory doesn't scrub mentions of it from older chats — if it came up in a conversation, that conversation still contains it.
• Delete the chat. In the sidebar, open the ⋯ menu next to the conversation and choose Delete. OpenAI says it leaves your chat history immediately and is scheduled for permanent deletion from its systems within 30 days. What it does not do is retract what ChatGPT already distilled from that chat into memory — that stays until you remove it there too.
• Wipe memory and switch it off. On the memory-summary page, the ⋯ menu has Delete and turn off memory. On the current system those are one action, not two: it clears what's shown and turns memory off, with no documented way to do only the first. It doesn't delete your past chats, and OpenAI notes that if you turn memory back on later, ChatGPT may create new memories from chats still in your history. OpenAI also says this control is "available on the web and rolling out on mobile soon," so it may not have reached the iPhone app yet. On the older saved-memories system the two stay separate: you can clear all entries and still leave memory running, or turn it off with Enable memory.
• Delete all chats. Settings → Data controls has options to archive or delete every conversation at once, including the ones inside projects. Combined with memory off, that's the clean slate. (If you've connected other apps to ChatGPT, disconnecting them stops future access but doesn't delete conversations that already used their data.)

For a one-off topic you don't want carried forward, start a Temporary Chat instead. Since 27 August 2026 you choose Personalized or Unpersonalized before the first message; either way, OpenAI's help article "Temporary chat in ChatGPT" (read 20 September 2026) says, "Neither option creates or updates memories while the chat remains temporary," and the chat stays out of your history unless you save it. (OpenAI's 12 June 2026 release note put it more simply: "Temporary Chats don't use existing memories or create new memories." The 27 August change is why that line is no longer the whole story.) And if your actual complaint is the reverse — ChatGPT keeps dragging old projects into new questions — see how to stop ChatGPT bringing up old conversations.

Why isn't ChatGPT saving my memories?

Work down this list, in rough order of likelihood:

• Memory is switched off. Check Settings → Personalization → Memory. On the older system the switch is labeled Enable memory.
• There isn't enough history yet. On the memory-summary system, OpenAI says the summary can sit empty if your account is new, you only just turned memory on, or there's too little chat history to draw from. It fills in as you use ChatGPT, and the summary's ⋯ menu has a Refresh option.
• Probably not "memory is full" — unless you're on the legacy list. If you're on the current memory summary, OpenAI's Memory FAQ documents no entry limit, no capacity figure and no "memory is full" state for it, so there's no published behaviour that would explain this — work through the rest of this list first. On the legacy saved-memories list it's likelier: OpenAI's release notes (entry dated 15 October 2025) do name a "memory full" state, and say automatic memory management "helps avoid" it rather than prevents it. See what OpenAI does and doesn't publish about memory capacity.
• You're in a Temporary Chat. A Temporary Chat doesn't create or update memories, and it stays out of your history unless you save it. Start a normal chat instead.
• The control you're looking for hasn't reached your device. The current memory system is still rolling out. OpenAI says of its Delete and turn off memory control, for instance, that it is "available on the web and rolling out on mobile soon" — so the iPhone app and the browser can show you different menus for the same account. Look at which controls your own Settings screen shows rather than assuming from someone else's screenshot.
• ChatGPT decided it wasn't worth keeping. It saves automatically only what it judges useful for future conversations. To force it, say so explicitly: "Remember that I'm the partner on the Acme engagement." OpenAI describes saved memories as exactly this — details you've directly told it to remember.

Does ChatGPT remember past conversations if you're not logged in?

No. You can use ChatGPT without an account at chatgpt.com, but everything above depends on having an account to attach memory and history to. Logged out, you get one conversation at a time, your chat history isn't saved, and the chat stays reachable only for the current session in the same browser. Memory and personalization aren't offered at all — personalization is one of the things OpenAI lists as unlocked by creating an account.

So nothing carries over to your next logged-out visit. And if you sign in or create an account partway through, only the conversation you're in at that moment is kept; earlier logged-out chats are gone. If you want ChatGPT to remember you, sign in, open Settings → Personalization → Memory, and switch memory on. Everything in this guide applies from there.

Does ChatGPT keep your conversations forever?

"Remembers" and "keeps" are different questions. What ChatGPT remembers is the memory described above. What it keeps is your chat history, and that follows your deletions rather than a clock:

• Chats you keep — active or archived — stay in your account until you delete them. There's no automatic expiry.
• Chats you delete vanish from your view immediately, and OpenAI's published guidance is that they're permanently deleted from its systems within 30 days — unless the chat was already de-identified and detached from your account, or OpenAI has to retain it for security or legal reasons. Deleted chats can't be recovered, including by support.
• Temporary Chats stay out of your history unless you save them, and the same help article (read 20 September 2026) says OpenAI "may keep a copy of a temporary chat for up to 30 days for safety purposes."
• Connected-app data used in a conversation stays stored with it; deleting the conversation is what removes it. Files you upload are kept separately — where Library is available they're saved there, and deleting the chat doesn't delete them; you remove them from Library.

Memory is a separate layer with its own retention: deleting chats doesn't clear what ChatGPT already distilled from them into the memory summary or saved-memories list — you remove that in the memory settings, as covered in the FAQ below. (OpenAI also says it may hold a log of deleted saved memories for up to 30 days for safety and debugging.)

ChatGPT memory, control and retention details above were re-verified on 2026-08-28 against OpenAI's own help center — the "Memory FAQ," "How to Delete and Archive Chats," and "Chat and File Retention Policies" articles, read in full — and the capacity statements re-verified on 2026-09-15 against a fourth source, the "ChatGPT — Release Notes" article (6825453), for the entries dated 15 October 2025 and 4 June 2026. help.openai.com refuses automated fetches with a 403, so the release notes were read through a text-extraction reader proxy and confirmed separately in a browser session, both on 2026-09-15. All quoted wording is OpenAI's. Where those pages are silent this guide says so rather than filling the gap: OpenAI publishes no absolute capacity figure on any plan, documents no entry limit or "memory is full" state for the current memory summary (its release notes do name one for the legacy saved-memories list), and documents no way to clear the summary without also turning memory off. OpenAI renames, moves and stages the rollout of these controls often — treat the labels as a guide and your own Settings screen as the authority for your account.

What "remembering" should mean for real work

If you want an assistant that remembers past conversations and the rest of your professional context, the bar is higher than chat memory. It needs three things ChatGPT's memory doesn't do:

For professionals, "remember" usually means a client, not a chat: the file a lawyer carries for years, the ledger an accountant reopens every quarter, the engagement a consultant picks up after a month away. We've written what an assistant can and can't hold for each — lawyers, accountants, consultants — with the rules each profession has to check first.

How Equerry approaches this

Equerry is a personal assistant built around exactly that boundary. You forward or CC the emails that matter to your own Equerry address, drop in a PDF or screenshot, dictate a voice memo, or save a note — and it remembers all of it together, not just what you typed into a chat. Then before your next client call it briefs you from what you've shared, and when you ask it something, it shows where the answer came from when it draws on your notes, emails or the web — tap through to the email or note.

Because you choose what it sees, there's no full-inbox handover and no passive capture — the threads you forward or CC are the threads it remembers. And every remembered fact is listed in one place you can review and delete, so you can audit what it thinks it knows about you in about thirty seconds. ChatGPT remembers for the conversation. Equerry is built to remember for the relationship — and show you exactly what it kept.

Last updated 2026-09-25.

All product names, logos, and brands are property of their respective owners. Equerry is not affiliated with, endorsed by, or sponsored by any other product mentioned; comparisons are provided for informational purposes.