Download attachments from a Google AI Studio chat
When you upload a PDF, a spreadsheet, a code file or a screenshot into AI Studio, the page shows a small preview. If you want the actual file back later, because you edited it in the conversation, because you lost the original, or because you want the chat and its files kept together, there is no download button on the page. Tools that copy what the page shows end up with a thumbnail, which is a picture of the file, not the file.
AI Studio Exporter downloads the real files. It is free for Markdown, JSON and plain text exports, attachments included.
#How to do it
- Install AI Studio Exporter.
- Open the conversation on aistudio.google.com.
- Click the extension's icon and click Export this conversation.
If the conversation has attachments, the download is a ZIP: the conversation document, and an attachments/ folder with every file in it. Links in the document point at those files, so extract the ZIP before you follow them.
How does the RPC auth work - 2026-07-31T06-40-38.zip
├── conversation.md
└── attachments/
├── network-log.har
└── screenshot of failing request.png
#Why the files are real
AI Studio keeps your attachments in your Google Drive and refers to them by an id. The exporter uses that id to download the file from Drive with your own signed-in session, at full size, and a file that AI Studio embeds directly in the conversation is saved as it is. So a PDF in your export is the PDF, a code file is the code file, and an image is the original image, not a preview. Nothing is sent through any server of ours: the file goes from Google to your browser to your disk.
#Google Docs, Sheets and Slides
A file created in Google's own editors is not a file you can download as it is, so the exporter converts it on the way out:
| In Drive | Saved as |
|---|---|
| Google Doc | Markdown (.md) |
| Google Sheet | CSV (.csv) |
| Google Slides | PDF (.pdf) |
| Google Drawing | PNG (.png) |
| Apps Script | JSON (.json) |
Any other Google-native file type falls back to a PDF.
#Choosing what to download
Two settings decide what is fetched:
- Images: download image attachments.
- Other attachments: download everything else.
Turn off the second one if you want the screenshots but not forty megabytes of CSV files. The choice is made by the file's real type, not its extension. Anything left out is counted, not forgotten: the message in the export still says, for example, 2 attachments not included in export. Turning both off skips the downloads entirely, which is the fastest way to export a very large conversation. The settings reference has the detail.
#Names, duplicates and limits
- Duplicate names. Two attachments both called
screenshot.pngare saved asscreenshot.pngandscreenshot (2).png. The number goes before the extension so each file still opens in the right program. - Long names. A name inside the archive is cut to 80 characters, keeping its extension, so the whole path stays well under Windows' 260-character limit (Windows Explorer refuses to open a ZIP with a path that long). The JSON export still records the full original name.
- Size. A single attachment over 100 MB is not downloaded, so one enormous file cannot exhaust the tab's memory. The export records the attachment, with its name, as not fetched.
- If a file cannot be fetched, the message keeps the file's name so the record stays honest, and the JSON export marks its path as
null.
#What about the fallback?
If AI Studio changes how it works and the normal route is unavailable, the extension falls back to reading the page. That cannot reach the files, only the previews the page has drawn, and the export tells you so with an amber badge on the toolbar icon. The badge guide explains it.
#Questions
#Will the files be the same quality as the original?
Yes, for ordinary files. They are downloaded unchanged. Google Docs, Sheets and Slides are the exception, because they have to be converted.
#Can I get the attachments without the conversation text?
The export always produces both, but you can ignore the document and use the attachments/ folder. For a whole library, the backup guide shows how the files are organised.
#Where is the conversation itself?
In conversation.md (or .pdf, .docx and so on, depending on the format you picked). See the Markdown guide.