Back to home

Guide

Does Claude have memory like ChatGPT?

Short answer: yes — with caveats

Claude has a memory feature that stores facts about you across chats, similar to ChatGPT's saved memory. Availability depends on your plan and which client you use (Claude.ai, Claude Desktop, Claude Code), and Anthropic has been rolling improvements out steadily.

Where Claude and ChatGPT differ most is in how you're expected to extend memory. ChatGPT leans on its persistent memory store. Claude leans on Custom Instructions, Projects, and — most powerfully — MCP connectors.

The three layers of Claude memory

  • Chat context. Everything in the current conversation. Resets when you start a new chat.
  • Custom Instructions. A stable profile string Claude sees every message. Best for tone, formatting, and persistent preferences.
  • Projects. Long-running workspaces that keep files and context scoped to a topic. Great for "everything about my company" or "everything about this book I'm writing."
  • MCP connectors. External data sources Claude reads and writes over the Model Context Protocol. This is where memory becomes essentially unlimited — the context lives in your data source, not in Claude.

Claude vs ChatGPT on memory

AspectChatGPTClaude
Persistent saved factsMemory feature, cappedMemory feature (varies by plan), plus Custom Instructions
Scoped contextProjectsProjects
Bring your own storeMCP connectorsMCP connectors

Both assistants land in roughly the same place: for anything beyond a handful of facts, MCP is the answer.

How to give Claude durable memory today

  1. Put stable preferences in Custom Instructions (Settings → Profile).
  2. Group topic-specific context into Projects.
  3. For everything else, use an MCP-connected notes workspace. Muninn is exactly that — a private notes store Claude reads and writes over MCP, so what you save is available in every future chat.

Related reading

Give Claude the long-term memory it doesn't have out of the box — sign up for Muninn.