OpenAI rolls out persistent file Library in ChatGPT

On March 23, 2026, OpenAI said on X that it is rolling out a new file Library inside ChatGPT. The linked ChatGPT release notes say that files you upload or create, including PDFs, spreadsheets, images, and other assets, are now saved automatically so they can be reused over time instead of being tied to a single conversation.

The release changes how file-based work happens in ChatGPT. Users can insert recent files directly from the composer, ask ChatGPT questions about previously saved documents, and browse everything from a new Library tab in the web sidebar. OpenAI also says generated images will continue to live in the Images tab, while mobile clients get recent-file access in the composer and file search support on iOS and Android.

• The full Library experience is available on web only.
• Files uploaded in Temporary Chat are not saved to the Library.
• The rollout is global for Plus, Pro, and Business users, with the EEA, Switzerland, and the UK coming later.

That sounds like a small UX change, but it materially changes ChatGPT from a chat-by-chat attachment tool into a more persistent workspace. Reusable file storage matters for research, document analysis, spreadsheet iteration, and multi-step drafting, where the friction of re-uploading files across sessions has been one of the biggest workflow breaks. It also gives ChatGPT a clearer memory boundary: files stay available until the user deletes them, but Temporary Chat remains separate.

The strategic signal is that leading AI assistants are becoming document workspaces rather than just prompt boxes. OpenAI is still limiting the full Library surface to web for now, but the split rollout already shows where the product is heading. The original X post is here, and the supporting release notes are on OpenAI Help Center.