Side by side, per November 2026
| Claude | ChatGPT | |
|---|---|---|
| What is stored | Individual memory topics, saved during chats | A memory summary, plus saved memories (legacy list available) |
| Past chats | Chat search: Claude searches old chats when you refer to them (paid plans) | Reference chat history: uses relevant information from past chats |
| Other sources | Project knowledge and memory | Custom instructions, files in Library, connected apps such as Gmail |
| Per project | Every Project has its own memory; “Use account memory” can cut it off | Projects use their own chats and files as context; Project-only memory is offered in some workspaces |
| Where to manage | Settings > Your Claude > Manage | Settings > Personalization > Memory |
| One chat without memory | Incognito, or switch Memory off via “+” before the first message | Temporary chat, choose Unpersonalized before you start |
| Team plans | Owner decides availability; owners cannot view members’ memories | Workspace settings decide; memories are not used for training by default |
How Claude does it
Claude saves memory as individual topics during chats, rather than summarizing conversations after they end. You can read and edit the topics under Settings > Your Claude, or tell Claude to remember or forget something. The structure is strict per Project: inside a Project, Claude uses that Project’s memory plus your account memory from chats outside Projects, but never the memory of other Projects. Sensitive topics (health, religion, politics, gender identity) are not stored unless you turn on “Include sensitive topics in memory”.
How ChatGPT does it
OpenAI’s help center describes Memory as drawing on several sources, depending on plan and region: past chats, saved memories, custom instructions, saved files in Library and content from connected apps. The improved memory keeps a broader summary that updates continually; you can switch back to the legacy list of saved memories. Below an answer, Sources can show which past chat, memory or file shaped it. Turning off Reference chat history schedules what was remembered from past chats for deletion within 30 days.
The real difference
The difference is less about how much each remembers and more about where the boundaries are. Claude draws a hard line around each Project, which suits work per client. ChatGPT reaches wider by default, into files and connected apps, which suits one person whose work spans everything. Neither keeps a full transcript: OpenAI says Memory does not retain every detail, and Claude stores selected topics.
The misunderstanding: switching tools moves the memory
You can import ChatGPT memory into Claude: run Anthropic’s import prompt in ChatGPT and paste the result into Claude (step by step). What arrives is a list of remembered facts. The chat history, the project files and what ChatGPT knew from connected apps stay behind. Anthropic calls the import experimental and says Claude focuses on work-related topics.
What this means for a team
Both memories are personal. On Claude Team, memory is built per member and owners cannot view or edit it. On ChatGPT Business, memory is part of each user’s workspace data. One notable difference sits next to memory: ChatGPT lets you share projects on every plan including Free, while Claude allows sharing Projects only on Team and Enterprise. Shared ChatGPT projects also let colleagues build on one another’s project chats; personal memories outside the project remain private. In Claude, a shared Project carries files and instructions, not what each person taught Claude. A team that uses both tools ends up with two memories per person, and neither knows what the other learned. For the Claude side in detail, see does Claude have memory.
Where Bombos fits
Claude and ChatGPT each remember for one person inside one tool. Bombos is a plugin that gives every AI tool a team uses, Claude, Claude Code, ChatGPT, Copilot, Grok, Cursor and your own agents, one shared memory with team knowledge, rules learned from corrections and what already lives in mail, documents and repos, so switching tools does not mean starting over. One person can try it first for their own agents.
Related questions
Can I move my ChatGPT memory to Claude?
Yes. Anthropic provides an import prompt you run in ChatGPT, then you paste the result under Settings > Your Claude > Start import. Chat history does not move.
Which one keeps work for different clients apart better?
Claude gives each Project its own memory by default, and you can cut a Project off from account memory with "Use account memory". ChatGPT projects use their own chats and files, and some workspaces offer a Project-only memory option.
Do Claude Team and ChatGPT Business share memory between colleagues?
No. In both, memory is built per user. Claude Team owners cannot even view members' memories.
Where do I manage memory in each?
Claude: Settings > Your Claude, then Manage next to Your memory. ChatGPT: Settings > Personalization > Memory.
