Understand the message
When ChatGPT indicates memory is full, it means the saved-memory store has reached its capacity and won't add new entries until you free up space. Existing memories aren't necessarily lost — some may be condensed.
ChatGPT memory limits
ChatGPT memory limits: how much can it remember?
ChatGPT's memory isn't bottomless. Saved memory has a finite capacity, so at some point it fills up — ChatGPT then stops adding new memories until you delete some, and it may condense older ones to save room. On top of that, the model's context window limits how much memory can be used in any single reply. OpenAI doesn't publish a fixed size, so here's how to think about it qualitatively.
Quick answer
There are two limits, and no published number
One limit is how much you can store overall — saved memory is finite and can fill up. The other is how much memory the model can read at once — bounded by the context window. OpenAI doesn't publish a fixed MB or token figure, so don't trust any specific number; manage memory qualitatively.
| Limit | What it means | What to do |
|---|---|---|
| Saved-memory capacity | Finite. When full, ChatGPT stops adding new memories and may condense existing ones. | Delete memories you no longer need to free up room. |
| Context window | The model can only read a finite amount of text per reply, so not all stored memory is used at once. | Keep important facts explicit and your list trimmed. |
| Reference chat history | Implicit, no visible list; its influence per reply is still bounded by the context window. | Toggle it off and delete old chats to reduce its reach. |
OpenAI doesn't publish a fixed memory size. Always confirm current limits and behavior on OpenAI's Memory FAQ.
When memory is full
What to do when ChatGPT says memory is full
A full memory store doesn't break ChatGPT — it just stops capturing new context. Here's how to manage it.
Open Manage memories
Go to Settings → Personalization → Memory and open "Manage memories" to see your stored facts. This is the only layer you can inspect and prune item by item.
Delete what's outdated
Remove entries that are stale, duplicated or no longer useful — or clear them all. Deleting memories lets ChatGPT resume saving new context. See delete memory.
Keep it curated
Treat memory like a maintained list, not an attic. Review it periodically so there's always room, keep sensitive data out, and use Temporary Chat for one-off tasks you don't want stored.
Related: how memory works · saved vs chat history · memory not working · delete memory
Misconceptions
What the limit is — and isn't
| Assumption | Reality |
|---|---|
| "Memory has a published size in MB." | OpenAI doesn't publish a fixed figure. Treat the limit qualitatively, not as a specific number. |
| "When full, ChatGPT forgets everything." | It mainly stops adding new memories; existing ones may be condensed rather than wiped. |
| "All my memory is used in every reply." | No — the context window bounds how much is injected at once. See how memory works. |
| "Chat history has a limit I can manage." | It's implicit with no visible list; reduce it by toggling off and deleting old chats. |
Outgrowing one assistant's memory
When one memory store isn't enough
If you keep hitting the limit, it's a sign you're funneling a lot of context into a single assistant. Spreading your work across models — and choosing which one handles which task — eases pressure on any one memory store, and lets you keep sensitive topics in memory-free sessions.
All-in-one
MultipleChat AI
multiplechat.aiUse ChatGPT, Claude, Gemini and other models in one place — choose where each prompt goes instead of feeding everything to a single assistant.
Privacy-first: MultipleChat doesn't save your chats to memory and doesn't share your data with model providers or let them train on it. Other privacy-focused options include Duck.ai, Proton Lumo, Brave Leo and Kagi Assistant; for full local control, self-hosted tools like Ollama or Jan keep everything on your device.
OpenAIChatGPT
chatgpt.comOffers Temporary Chat (no history, not used for training), a training opt-out under Settings → Data Controls, and editable memory. By default, consumer chats may be used to improve models unless you opt out — verify on OpenAI's help center.
AnthropicClaude
claude.aiAnthropic gives you data and training controls and has historically taken a privacy-conscious stance on consumer chats, but its training and retention settings have changed over time — check your current privacy settings in the app.
GoogleGemini
gemini.google.comGemini Apps Activity lets you pause history and delete past activity, but sampled conversations can be reviewed by humans, so avoid sharing confidential information. Review your Activity controls.
MicrosoftCopilot
copilot.microsoft.comMicrosoft 365 Copilot (business) doesn't use your organization's data to train models and adds Enterprise Data Protection; consumer Copilot has a privacy dashboard to review and delete your history. Check current Microsoft policies.
PerplexityPerplexity
perplexity.aiPerplexity has an "AI data retention" setting you can switch off so your searches aren't used to improve its models, plus account-level data controls. Confirm the current options in settings.
All guides
ChatGPT memory: every guide
How memory worksThe technical walkthrough.
Saved vs chat historyThe two layers compared.
Turn memory on/offManage the settings.
Delete memoryClear individual or all.
Memory & privacyRisks and how to reduce them.
Memory & trainingIs it used to train models?
Temporary ChatMemory-free conversations.
Memory not working?Troubleshooting steps.
Memory limitsHow much it can hold.
Team & EnterpriseMemory and data controls.
Memory vs custom instructionsWhat's the difference?
Memory FAQEvery common question.
FAQ
ChatGPT memory limits — answers
How much it can hold, what happens when it's full, and how to manage capacity.
Is there a limit to how much ChatGPT can remember?
Yes. Saved memory has a finite capacity. When your memory is full, ChatGPT stops adding new memories until you delete some, and it may consolidate or condense existing memories to make room. OpenAI doesn't publish a fixed size figure, so it's best to think of it qualitatively: there's a ceiling, and you manage it by deleting memories you no longer need. Check OpenAI's help center for current behavior.
How many memories can ChatGPT store?
OpenAI doesn't publish a specific number of memories or a fixed storage size, so we won't invent one. What is known is that the saved-memories store is finite: at some point it reports that memory is full and stops adding new entries until you free up space. Treat the limit as real but unspecified, and manage it by reviewing and deleting memories.
What happens when ChatGPT memory is full?
When memory is full, ChatGPT stops automatically saving new memories until you delete some existing ones. It may also consolidate or condense memories so related facts take up less room. If you want ChatGPT to keep capturing new context, open Manage memories and remove entries you no longer need. Verify the exact behavior on OpenAI's help center.
Does ChatGPT delete old memories automatically?
Rather than simply deleting, ChatGPT may consolidate or condense memories to manage capacity, combining related facts so they take less space. It doesn't guarantee that any specific older memory is kept forever. To be sure of what's stored, review the saved-memories list yourself and delete what you don't want. Behavior can change, so confirm on OpenAI's help center.
How big is ChatGPT's memory in MB or tokens?
OpenAI doesn't publish a fixed MB or token figure for saved memory, so any specific number you see is likely a guess. We recommend treating the limit qualitatively: it's finite, it can fill up, and you manage it by deleting memories. If you need a precise figure, the only reliable source is OpenAI's own documentation.
Does the context window limit memory too?
Yes, separately. Even if you have plenty stored, the model can only read a finite amount of text at once — its context window. Memory that's injected into a reply has to fit within that budget alongside your prompt and the conversation. So there are two limits at play: how much you can store overall, and how much can be used in any single reply.
Why won't ChatGPT save a new memory?
The most common reason is that memory is full, so ChatGPT has stopped adding new entries until you delete some. Other possibilities are that memory is turned off, or that the system didn't judge the detail worth saving. Open Manage memories to check whether it's full and to free up space. If problems persist, see our memory-not-working guide.
How do I free up memory in ChatGPT?
Go to Settings → Personalization → Memory and open Manage memories. Delete individual entries you no longer need, or clear them all. Removing memories lets ChatGPT resume saving new ones. You can also tell ChatGPT to forget something directly in chat. The interface changes, so check OpenAI's help center for the current steps.
Does reference chat history have a limit?
Reference chat history is an implicit layer with no visible list, so there's no published, manageable size you can inspect. In practice, how much of your past chats can influence any reply is still bounded by the context window. To reduce its influence, turn the toggle off and delete old conversations you don't want drawn on.
Will hitting the limit make ChatGPT forget what it already knows?
Reaching the cap mainly stops new memories from being added; it doesn't necessarily erase what's already stored, though ChatGPT may condense related memories to save space. If you want to be certain a specific fact stays, keep your list trimmed so there's room, and verify it's present in Manage memories. Behavior can change, so confirm on OpenAI's help center.
How should I manage memory so I don't hit the limit?
Review your saved memories periodically and delete entries that are outdated or no longer useful, so there's always room for new context. Keep truly sensitive data out of memory, and use Temporary Chat for one-off tasks you don't want stored. Treating memory like a curated list, rather than letting it grow unchecked, keeps it useful and within its limits.