Are these the same as a loose chat with memory on?

No. Both group instructions and files for one body of work. A loose chat does not automatically inherit the project’s rules and sources. Tutorials claude-projects-primer-flujo and chatgpt-projects-primer-flujo walk through building the container.

Can I run a Project in Claude and another in ChatGPT for the same client?

Yes, and sometimes that helps when teams split across claude.ai and chatgpt.com. Align instructions and filenames in both and avoid putting sensitive data on accounts you do not control. Do not duplicate the same thread in two places without a clear owner.

Which shares better with teammates?

ChatGPT offers invites and links on more consumer and small-team plans. Claude focuses org sharing on Team and Enterprise. Check view vs edit permissions before uploading contracts.

What about memory across projects?

On ChatGPT, project-only memory keeps context inside the Project; shared ones use that mode. On Claude, each project has its own memory and you can limit use of account memory inside the project. Neither replaces a wiki — they are chat containers.

What if a ChatGPT Project already has many threads and stale files?

You do not have to delete it: split stable instructions from operating rules, remove obsolete sources and name chats by phase. Flow chatgpt-projects-organizar-trabajo targets that, after chatgpt-projects-primer-flujo.

Claude Projects vs ChatGPT Projects

Two containers with instructions, files and grouped chats: one on claude.ai built around shared knowledge and writing; one on chatgpt.com with project memory and broad sharing. Which to set up for a client, a report or work that runs for weeks.

	Claude	ChatGPT
Best for	Repeated briefs, reports and docs with instructions + reference files on claude.ai	Ongoing work on chatgpt.com: several chats, files, search or GPTs in the same habit
Instructions and sources	Project instructions + knowledge base (uploads or Drive); apply to each new chat inside the project	Instructions in Project settings + files or pasted text; override global custom instructions while you work here
Context across project chats	What you add to project knowledge yes; one chat thread is not shared with another unless it lives in that base	Project memory: chats can use other threads and files in the same Project (default or project-only)
Memory and context limits	Per-project memory; option to separate from account memory; automatic RAG on paid plans as knowledge grows	Choose default or project-only memory at creation (changeable later); shared projects stay project-only
Sharing and teams	Sharing on Team and Enterprise plans (org visibility or invitees with view/edit)	Sharing on supported plans (including Free per workspace); chat or edit access; member memory isolated
Long-term organization	Move chats into the project, archive finished projects; free tier caps how many projects you can create	Move chats into the project, weekly file and instruction hygiene; file limits depend on plan
Main risk	Assuming one chat «remembers» another thread without moving that into project knowledge	Blending global memory with a sensitive case or leaving key threads outside the Project

Verdict

Pick Claude Projects when your team already works on claude.ai, the deliverable is long-form text or documents with steady tone and you want a clear knowledge panel (instructions + files that apply to every chat in the project). On paid plans, RAG mode expands knowledge when file volume grows. Pick ChatGPT Projects when you need several threads that keep drawing on the same container’s history, sharing the project with people outside an Anthropic Enterprise workspace (Free plans and up, per settings) or bounding memory to the project with project-only memory — useful for mixed teams or sensitive work. They are not mutually exclusive: many teams keep one Project per tool and per case. First container: claude-projects-primer-flujo and chatgpt-projects-primer-flujo; if a ChatGPT Project already runs for weeks, chatgpt-projects-organizar-trabajo.