Feature Request — Storage: Large/Bulk Download & Archive

ChatGPT Feature requests
, , ,
1

Feature Request — Storage: Large/Bulk Download & Archive

With the introduction of the larger ChatGPT Library and storage management, I think we are reaching the point where bulk file management needs to scale with the amount of storage offered.

I currently have approximately 40 GB of my 100 GB storage allowance occupied by images.

As one small data point, I sampled my 100 most recent Library images:

  • 100 images: ~205.5 MiB

  • Average: ~2.06 MiB per image

  • 92% were ChatGPT-generated images

  • All 100 in the sample were PNG files

If that sample is even approximately representative, 40 GB of images would be on the order of ~19,000 individual files.

At that scale, downloading files through ordinary Library multi-selection stops being practical.

Requested feature: Large / Bulk Download

I would love to see a proper bulk export/download mode for Library and Images, designed for thousands or tens of thousands of assets.

Ideally this could include:

  • Select all matching files, not only the files currently visible or manually selected.

  • Bulk selection by:

    • generated vs uploaded

    • file type

    • date/date range

    • conversation

    • project

    • folder

    • search/filter results

  • Download entire Image Library

  • Download entire Library

  • Export selected content as ZIP archives, preferably automatically split into manageable chunks for very large collections.

  • A manifest such as manifest.json or CSV containing useful metadata:

    • filename / asset ID

    • creation date

    • file size

    • MIME type

    • image dimensions

    • generated/uploaded status

    • originating conversation/project where available

    • generation metadata where available

  • Resumable downloads or individually downloadable archive parts.

  • Checksums for large exports.

  • An “Export, then delete” workflow for storage cleanup, with deletion occurring only after the archive has been successfully prepared.

Storage management would benefit from this too

The new storage meter makes it much easier to see when Library usage is becoming substantial, but once someone has accumulated tens of gigabytes, the obvious next question is:

“How do I archive this locally without clicking through thousands of files?”

A storage dashboard could therefore also expose breakdowns such as:

Images — 40 GB
Documents — X GB
Generated files — X GB
Uploaded files — X GB

with an Export action beside each category.

Why not just use Account Data Export?

Account Data Export solves a different problem.

It is useful for obtaining a copy of account data, but a user who simply wants to archive 20–50 GB of generated images should not need to request a complete account export.

Library already behaves increasingly like persistent user storage, so having a storage-oriented archive/download workflow directly inside Library would make much more sense.

Example UX

Something as simple as:

Library → Storage → Images → Export

Then:

Export 18,742 files — 39.8 GB

Format: Original files
Include metadata manifest: ✓
Split archives into: 5 GB parts

Prepare Download

Once complete, ChatGPT could provide the archive parts for download.

For very large libraries, preparing the archive server-side and notifying the user when it is ready would also avoid keeping the browser session alive throughout the operation.

Why I think this matters

100 GB is large enough that ChatGPT Library is no longer just a collection of occasional attachments.

For users who generate images frequently, work with datasets, create documents, or use ChatGPT as part of a creative workflow, it can become a substantial personal archive.

Giving users large-scale export, backup, and cleanup tools proportional to the storage capacity offered would make Library much easier to trust and maintain long-term.

TL;DR: Please add a true Library-scale bulk download/export system — including “select all matching,” archive splitting, metadata manifests, and ideally an export-and-delete workflow — so tens of gigabytes of generated content can be backed up without manually selecting thousands of individual files.

2

@DysTopia and I had a chat about this, and I think we have a workaround until these file organization and management features are implemented.

In short:

Use the desktop app and, in a new chat, ask it to go through all chats, or only those up to a certain date, and collect all the images. The agent will happily oblige and do the manual work for you.

For older images, a full data export via privacy.openai.com is the way forward. You can then ask ChatGPT/Codex to work through the archive and collect all the images.

There is still a lot of headroom for improvement when it comes to token savings and storing the metadata, but that’s the gist of it.

I hope this helps someone else looking for a similar solution.

3

Exactly, the idea is to make the download aspect of the files as user friendly as possible to preserve structures set by the user for workflows and productivity.

In my case images always have a raw input file, an interpreted prompt file and a markdown file to document the image provenance and metadata.

Or… here me out! The GPT Image *.gpti file format is introduced with such things included. After all most file formats are a sort of container :grin: