Set a fixed baseline
Write your stable preferences into custom instructions — your role, the tone you want, formatting rules. These are always applied, so they're the dependable foundation for every chat.
Memory vs custom instructions
ChatGPT memory vs custom instructions.
They both make ChatGPT feel more personal, and they're easy to confuse — but they work differently. Memory is auto-managed: ChatGPT writes and grows it for you. Custom instructions are hand-written: you set them and they stay fixed until you edit them. This guide explains the difference and how to use the two together.
Quick answer
Auto-managed vs hand-written
Memory builds up on its own and changes as you chat. Custom instructions are text you author and that stays put until you rewrite it. One is adaptive; the other is stable. They're separate features and you can run them at the same time.
| Memory | Custom instructions | |
|---|---|---|
| Who writes it | ChatGPT, automatically (or when you say "remember this"). | You, in your own words. |
| How it changes | Grows and updates over time. | Fixed until you edit it. |
| Applied when | When retrieval judges a memory relevant to your prompt. | Always applied. |
| Predictability | Adaptive, less predictable. | Consistent and predictable. |
| Best for | Evolving context that accumulates naturally. | Stable preferences like tone and format. |
Both are personalization features and can be used together. Confirm the current settings layout on OpenAI's Memory FAQ.
How to combine them
Using memory and custom instructions together
A practical setup: lock in your stable preferences with custom instructions, and let memory handle the evolving details.
Let memory adapt
As you work, let saved memories capture evolving context — current projects, recurring topics, things you say "remember this" about. This layer grows without you maintaining it.
Audit memory periodically
Because memory is auto-managed, review the saved-memories list now and then. Edit anything inaccurate and delete anything you'd rather keep out — your custom instructions stay untouched.
Keep sensitive items deliberate
For sensitive preferences, prefer custom instructions where you control the exact wording, and use Temporary Chat for private one-off tasks so nothing is saved either way.
Related: how memory works · saved vs chat history · memory limits · privacy & data
Common confusions
What people mix up
| Assumption | Reality |
|---|---|
| "They're the same feature." | No — separate features. Memory is auto-managed; custom instructions are hand-written and always applied. |
| "Editing custom instructions changes my memory." | It doesn't. They're stored and managed independently. |
| "Turning off memory clears custom instructions." | No — custom instructions stay in effect until you remove them yourself. |
| "Memory only stores what I explicitly add." | Memory is auto-managed and can save things on its own. See the two layers. |
Personalization, your way
Decide what each model knows about you
Memory and custom instructions both shape one assistant. If you use several AI tools, each stores its own context separately — and you may not want all of it concentrated in one place. A multi-model workspace lets you choose which model sees which prompt while keeping sensitive topics out of any persistent personalization.
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
Memory vs custom instructions — answers
How the two features differ, how they reach the model, and how to combine them.
What is the difference between ChatGPT memory and custom instructions?
Memory is auto-managed: ChatGPT writes and updates saved memories about you over time, and you can edit or delete them. Custom instructions are manual: you write your own answers to "what should ChatGPT know about you" and "how should it respond," and they stay fixed until you change them. Memory grows on its own; custom instructions are user-authored and stable. They're separate features you can use together.
What are custom instructions in ChatGPT?
Custom instructions are a personalization feature where you state, in your own words, what ChatGPT should know about you and how it should respond — for example your profession, the tone you prefer, or formatting rules. They're always applied to your chats, are written by you, and are not auto-updated. You change them only by editing them yourself.
Is ChatGPT memory the same as custom instructions?
No. They're different features. Saved memories behave a bit like custom instructions because they shape responses, but ChatGPT manages and grows them automatically, while custom instructions are fixed text you author yourself. Memory can change without you doing anything; custom instructions only change when you edit them.
Can I use memory and custom instructions together?
Yes. They're complementary. Custom instructions are good for stable preferences you want applied every time, such as tone or format. Memory is good for evolving context that builds up naturally as you chat. Using both gives you a fixed baseline plus an adaptive layer. Verify current behavior on OpenAI's help center.
Which is more reliable for getting consistent answers?
Custom instructions are more predictable because they're always applied and only change when you edit them. Memory is more adaptive but less predictable, since it's auto-managed and retrieval decides what's relevant to a given prompt. If you need a preference applied consistently, putting it in custom instructions is the more dependable choice.
Do both reach the model the same way?
Conceptually yes: both add context to the model at inference time rather than retraining it. Custom instructions and relevant memory are provided to the model alongside your prompt, like a hidden system note, so the model conditions its reply on them. The model's base weights are unchanged. The main difference is who writes the content and whether it updates automatically.
How do I edit custom instructions and memory?
Custom instructions are edited in your settings under the personalization area, where you rewrite the text yourself. Saved memories are managed in Settings → Personalization → Memory, where you can view, edit or delete each stored fact. The interfaces change, so check OpenAI's help center for the current location of each.
If I delete a memory, does it affect custom instructions?
No. They're stored separately. Deleting a saved memory removes that auto-managed fact but leaves your custom instructions untouched, and editing custom instructions doesn't change your saved memories. You manage each one independently.
Which should I use for sensitive preferences?
If a preference is sensitive, custom instructions give you direct control over the exact wording and you can remove it at any time, whereas memory may capture things automatically. Either way, avoid putting truly sensitive data (financial, medical, credentials) into persistent personalization, and use Temporary Chat for private one-off tasks.
Does turning off memory disable custom instructions?
No. Memory and custom instructions are separate settings. Turning off saved memories and reference chat history stops the auto-managed layer, but your custom instructions remain in effect unless you clear them. To stop both, you'd disable memory and also remove or clear your custom instructions.
Why does ChatGPT remember a preference I never put in custom instructions?
Because memory is auto-managed. ChatGPT can save a preference to memory or reflect it through reference chat history even if you never wrote it into custom instructions. If you want to remove it, check and prune your saved-memories list and, if needed, turn off reference chat history.