ChatGPT Projects vs Memory vs Custom Instructions: Complete Guide (2026)

Q: What is the difference between ChatGPT Memory and Custom Instructions?

ChatGPT Memory automatically learns from your conversations and stores up to ~1,500 words of facts about you. Custom Instructions are a static text block (1,500 characters per field) that you manually set once. Memory updates dynamically; Custom Instructions stay the same until you change them.

Q: What are ChatGPT Projects and how do they differ from Memory?

ChatGPT Projects are organized folders that combine conversations, files, and custom instructions under a single workspace. Unlike Memory, which is global and automatic, Projects give you per-project context with file uploads and project-specific instructions. Projects are available on Plus, Team, and Pro plans.

Q: Can I use ChatGPT Memory, Custom Instructions, and Projects together?

Yes. Memory provides global preferences that apply everywhere, Custom Instructions set your default behavior, and Projects add project-specific context on top. They work in layers: Custom Instructions form the base, Memory adds learned facts, and Projects overlay project-specific instructions and files.

Q: Which is better: ChatGPT Memory or Custom Instructions?

Neither is strictly better — they serve different purposes. Use Custom Instructions for static, always-relevant information (your role, preferred format, tone). Use Memory for dynamic facts that ChatGPT learns over time. For best results, use both together.

Q: Are ChatGPT Projects available on the free plan?

No. ChatGPT Projects are only available on paid plans: Plus ($20/month), Team, Pro ($200/month), and Enterprise. Free plan users only have access to Memory and Custom Instructions.

ChatGPT now offers three different personalization features— Memory, Custom Instructions, and Projects — and most users are confused about how they differ, when to use each one, and how they work together. This guide breaks down all three features with clear examples, a side-by-side comparison table, and practical advice for getting the most out of ChatGPT's personalization system.

Last updated: May 2, 2026 · 15 min read

TL;DR

Custom Instructions = static text you set once (1,500 chars per field). Memory = auto-learned facts from conversations (~1,500 words). Projects = organized folders with files + custom instructions (paid plans only). They work in layers together. But all three are trapped inside ChatGPT — for cross-platform memory, use AI Memory.

What Is ChatGPT Memory?

ChatGPT Memory is an automatic learning system that lets ChatGPT remember key facts about you across conversations. When enabled, ChatGPT analyzes your chats in real time, identifies important details — your name, preferences, project context, technical stack, communication style — and saves them as short memory entries.

These memories are then injected into future conversations so ChatGPT can provide personalized responses without you having to repeat yourself. For example, if you mention you're building a React app with TypeScript, ChatGPT will remember that and give you TypeScript-specific code suggestions in future coding discussions.

Key Characteristics of ChatGPT Memory

Automatic:ChatGPT learns and saves memories without you explicitly telling it to. You can also prompt it to remember specific things with phrases like "remember this."
~1,500 word limit: The total storage capacity is approximately 1,500 words, which translates to roughly 200 individual memory entries. Once full, older memories are automatically deleted.
Toggle on/off: You can enable or disable memory entirely in Settings → Personalization → Memory. You can also delete individual memories.
Global scope: Memories apply across all your conversations (except those in temporary/incognito chats).
No file support: Memory only stores text snippets. You cannot attach files, documents, or images to memories.

💡 When Memory Works Best

Memory is ideal for lightweight personal preferences that apply everywhere: your name, role, preferred coding language, communication tone, and recurring project details. Think of it as ChatGPT getting to know you over time. For a deep dive, see our complete ChatGPT memory guide.

What Are ChatGPT Custom Instructions?

Custom Instructions are a static text blockwhere you tell ChatGPT how to behave and what context to consider in every conversation. Unlike Memory, Custom Instructions don't change automatically — you set them once and they remain until you manually edit them.

Custom Instructions consist of two fields:

"What would you like ChatGPT to know about you?" — Your background, role, preferences, and context. (1,500 character limit)
"How would you like ChatGPT to respond?" — Response format, tone, length, and style preferences. (1,500 character limit)

Key Characteristics of Custom Instructions

Manual & static:You set the content yourself. It doesn't change unless you edit it.
1,500 characters per field:That's approximately 250-300 words per field, or 500-600 words total across both fields.
Global scope: Custom Instructions apply to every conversation, including those with GPTs (custom versions of ChatGPT).
No file support: You can only enter plain text. No file uploads.
Available on all plans: Free, Plus, Team, Pro, and Enterprise all support Custom Instructions.

💡 When Custom Instructions Work Best

Custom Instructions are perfect for permanent context that never changes: your job title, preferred response format (bullet points vs. paragraphs), always-on constraints (e.g., "always use TypeScript"), and communication preferences. Think of it as setting default parameters. For examples, see our Custom Instructions guide.

What Are ChatGPT Projects?

ChatGPT Projects are organized workspaces that group related conversations together with dedicated files and custom instructions. Think of them as folders where each folder has its own context, documents, and chat history — separate from your main ChatGPT conversations.

Each Project can contain:

Project-specific custom instructions: Set instructions that only apply within this project, separate from your global Custom Instructions.
File uploads: Upload documents, code files, spreadsheets, images, and other files directly to the project. ChatGPT can reference these files across all conversations in the project.
Organized conversations: All chats within a project are grouped together, making it easy to find and continue related discussions.
Persistent project memory: Context from previous conversations within the project carries over to new chats in the same project.

Key Characteristics of ChatGPT Projects

Folder-based organization: Projects act as containers for related work — software development, content creation, research, data analysis.
File support: Unlike Memory and Custom Instructions, Projects let you upload files that ChatGPT can reference across conversations.
Per-project instructions: Each project has its own custom instructions on top of your global Custom Instructions.
Paid plans only: Projects are available on ChatGPT Plus ($20/month), Team, Pro ($200/month), and Enterprise. Not available on the free plan.
No storage limit:Projects don't count toward your Memory storage. Files and project instructions have their own limits.

⚠️ Projects Are Paywalled

If you're on the free ChatGPT plan, you cannot use Projects at all. You need at least ChatGPT Plus ($20/month) to access this feature. Memory and Custom Instructions, by contrast, are available on all plans including free.

ChatGPT Projects vs Memory vs Custom Instructions: Side-by-Side Comparison

Here's a detailed comparison of all three personalization features across key dimensions:

Feature	Memory	Custom Instructions	Projects
Scope	Global (all chats)	Global (all chats)	Per-project only
Persistence	Auto-learned, can be auto-deleted	Static until manually edited	Persistent within project
How it's set	Automatic (learned from conversations)	Manual (you write it)	Manual (instructions + files)
Storage limit	~1,500 words (~200 entries)	1,500 chars per field (2 fields)	Per-project instructions + file uploads
File support	❌ No	❌ No	✅ Yes (documents, code, images)
Organization	Flat list of memories	Single text block	Folder-based with grouped chats
Availability	All plans (Free, Plus, Team, Pro)	All plans (Free, Plus, Team, Pro)	Paid only (Plus, Team, Pro)
Cross-platform	❌ ChatGPT only	❌ ChatGPT only	❌ ChatGPT only
Best for	Personal preferences & recurring context	Default behavior & static context	Multi-file projects & organized workflows

When to Use Each Feature: Practical Use Cases

Understanding when to use each feature is key to getting the best results from ChatGPT. Here's a practical breakdown:

Use Memory When...

You want ChatGPT to automatically learn your preferences over time
You have lightweight facts that apply to all conversations (name, role, coding language)
You want zero maintenance — set it and forget it
You're on the free plan and can't use Projects

Use Custom Instructions When...

You have permanent context that should never change (job title, always-on constraints)
You want to control response format (e.g., "always respond in bullet points")
You need information to be available immediately — not learned over time
You want to set tone and style defaults for every conversation

Use Projects When...

You're working on a complex, multi-file project (software development, research, writing)
You need project-specific context that shouldn't leak into other conversations
You want to upload reference files (codebases, documents, data)
You need to organize related conversations in one place

The Power Combo: Using All Three Together

The best approach is to use all three features in layers:

Custom Instructions as the base layer: set your role, preferred format, and always-on constraints.
Memory as the learning layer: let ChatGPT pick up new preferences, project details, and context automatically.
Projects as the workspace layer: create dedicated projects for complex, multi-file work that needs its own context and organization.

This way, ChatGPT has a solid foundation (Custom Instructions), continuously improves (Memory), and has deep context for specific work (Projects).

The Limitation: Everything Is Trapped in ChatGPT

Here's the uncomfortable truth about all three features: everything is locked inside ChatGPT. Your memories, custom instructions, and project files exist only within OpenAI's ecosystem. This creates several problems:

No cross-platform portability: If you also use Claude, Gemini, DeepSeek, or other AI tools, none of your ChatGPT personalization carries over. You start from scratch on each platform.
No export for memories:You can export your conversations, but there's no built-in way to export your saved memories or custom instructions.
Auto-deletion risk: ChatGPT memories are silently deleted when the limit is reached. Important context can vanish without warning.
Vendor lock-in:All your personalization data lives on OpenAI's servers. If you switch AI tools, you lose everything.
No search across conversations:You can't do full-text search across your entire conversation history. The sidebar search only matches titles.

🔒 The Vendor Lock-In Problem

You've invested hundreds of hours building context in ChatGPT — memories, custom instructions, organized projects. But all of that data is trapped. If you want to try Claude for coding or Gemini for research, you lose all that personalization. That's not a workflow — that's a lock-in.

How AI Memory Solves the Cross-Platform Problem

AI Memoryis a free tool that breaks your ChatGPT memories out of their silo and makes them work across every AI platform. Here's how it complements — and extends — all three ChatGPT personalization features:

Cross-platform memory: Store and access your AI context across ChatGPT, Claude, Gemini, and DeepSeek. Set up your preferences once, use them everywhere.
Unlimited storage: No 1,500-word cap. Store millions of words of context across thousands of conversations.
Full-text search: Search across every message in every conversation — not just short memory snippets. Find exact details from months ago in seconds.
Export & import: Export your ChatGPT data, import it into AI Memory, and search it forever. No more losing data to auto-deletion.
Memory injection: AI Memory can automatically inject relevant context into any AI chat, giving you personalized responses on any platform.
Data ownership: Your data stays on your device. You control it completely — no vendor lock-in.

ChatGPT Personalization vs AI Memory: Feature Comparison

Here's how ChatGPT's built-in features compare to AI Memory as an external memory solution:

Feature	ChatGPT (Memory + CI + Projects)	AI Memory
Storage	~1,500 words (Memory) + 3,000 chars (CI)	Unlimited
Full-Text Search	❌ Title-only search	✅ Searches every message
Cross-Platform	❌ ChatGPT only	✅ ChatGPT, Claude, DeepSeek, Gemini
Export / Import	⚠️ Conversations only (no memory export)	✅ JSON, CSV, Markdown
Auto-Deletion	❌ Yes (memories silently removed)	✅ Never — permanent storage
Memory Injection	Automatic only (within ChatGPT)	✅ Into any AI chat on any platform
Data Ownership	OpenAI's servers	✅ Your device / your server
Price	Free (Memory/CI) / $20+ (Projects)	✅ Free (Chrome extension + web app)

Take Your AI Memory Beyond ChatGPT

Try AI Memory — Free, Unlimited, Cross-Platform

Don't let your ChatGPT memories, custom instructions, and project context stay locked in one platform. With AI Memory, you can:

✅ Search across all your AI conversations — ChatGPT, Claude, DeepSeek, Gemini
✅ Store unlimited memories — no 1,500-word cap
✅ Export & import freely — JSON, CSV, Markdown
✅ Inject context into any AI chat — automatic memory injection
✅ Own your data — stored locally, fully under your control

Try AI Memory Free →

Free forever. No account required. Works with any ChatGPT data export.

Frequently Asked Questions

What is the difference between ChatGPT Memory and Custom Instructions?

ChatGPT Memory automatically learns from your conversations and stores up to ~1,500 words of facts about you. Custom Instructions are a static text block (1,500 characters per field) that you manually set once. Memory updates dynamically as you chat; Custom Instructions stay the same until you change them.

What are ChatGPT Projects and how do they differ from Memory?

ChatGPT Projects are organized folders that combine conversations, file uploads, and project-specific custom instructions. Unlike Memory — which is global and automatic — Projects give you per-project context with file support. Projects are available on Plus, Team, Pro, and Enterprise plans only.

Can I use ChatGPT Memory, Custom Instructions, and Projects together?

Yes. They work in complementary layers: Custom Instructions form the base (default behavior), Memory adds learned facts that apply globally, and Projects overlay project-specific instructions and file context. Using all three together gives ChatGPT the richest understanding of your needs.

Which is better: ChatGPT Memory or Custom Instructions?

Neither is strictly better — they serve different purposes. Use Custom Instructions for static, always-relevant information (your role, preferred format, tone). Use Memory for dynamic facts that ChatGPT learns over time (project details, evolving preferences). For best results, use both together.

Are ChatGPT Projects available on the free plan?

No. ChatGPT Projects are only available on paid plans: Plus ($20/month), Team, Pro ($200/month), and Enterprise. Free plan users only have access to Memory and Custom Instructions.

Is there a way to use ChatGPT memory across platforms like Claude or Gemini?

ChatGPT memory is locked to ChatGPT and cannot be used with other AI platforms. For cross-platform memory, you can use AI Memory, which stores and searches your conversations across ChatGPT, Claude, DeepSeek, and Gemini with unlimited storage and memory injection.