Can I upload a ZIP file to ChatGPT?

Yes, the upload goes through. Reading it is a separate step: ChatGPT's file-analysis tools have to unpack the archive before it sees any file inside. ZIP is not on OpenAI's list of supported file types. OpenAI's support team wrote on its community forum in September 2026 that the file-analysis tools may not be able to unpack or inspect the archive, even when the upload appears to succeed.

Why does ChatGPT say it can't open my ZIP?

Because unpacking runs in ChatGPT's file-analysis tools, and when they error the archive stays closed. Forum reports from 2024 to September 2026 describe ZIPs that uploaded fine and then could not be read, including inside Projects and Business workspaces. A plain text file needs no unpacking, so the same contents combined into one file avoid the step that fails.

What does OpenAI recommend instead?

Its support reply suggests extracting the ZIP on your computer and uploading the individual files. Each of those files counts toward the upload quota: 80 files every 3 hours, and 3 a day on the Free plan, as of September 2026. In our sample of 60 public code repositories, 51 held more than 80 files. Combined into one text file, the whole archive spends one upload.

Can Claude read ZIP files?

ZIP is not on the list of document types Anthropic says Claude works with (PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB and JSON). Its help pages say nothing about archives. A chat also takes at most 20 files, and 59 of the 60 repositories we measured held more than that. One combined TXT file is on the list and takes one of the 20.

Can Gemini read ZIP files?

Yes, Google lists ZIP as supported, with limits: a ZIP can contain up to 10 files, can be up to 100 MB and can't include video or audio. Every one of the 60 repositories we measured held more than 10 files, the median 183. For code, Gemini's Import Code takes one folder or GitHub repository of up to 5,000 files instead. For documents, a combined text file carries everything in one.

Which archives does FileConcat open?

ZIP, TAR, TAR.GZ (or .tgz), GZ, 7Z, RAR, BZ2 and XZ, all unpacked in your browser. The documents inside, PDF, Word, Excel and PowerPoint included, are read as text the same way loose files are. A password-protected or damaged archive won't open; unpack that one on your computer and drop the folder.

Are my files uploaded to a server?

Not to us. FileConcat unpacks the archive in your browser, and there is no account to create. The only upload is the one combined file you choose to send to ChatGPT.

Runs in your browser. Files not uploaded to us.

You can upload a ZIP to ChatGPT. Reading it is the part that can fail.

The upload goes through, then ChatGPT has to unpack the archive before it sees a single file inside. ZIP is not on OpenAI's list of supported file types, and OpenAI's support team says the unpacking may not work even when the upload looks fine. Drop the ZIP here instead: everything inside is read in your browser and comes back as one text file, with nothing left to unpack.

.zip, .7z, .rar, .tar and .tar.gz unpacked here
PDFs and Office docs inside read too
No sign-up, files not uploaded to us

Drop your ZIP here

Unpacked and read in your browser.

Can ChatGPT, Claude and Gemini read a ZIP?

Only Gemini documents ZIP support, and it caps what an archive may hold.

Where	Listed as supported	What gets read
ChatGPT	No	The upload goes through, but the files inside are read only if its file-analysis tools unpack it, and OpenAI support says that may fail.
Claude	No	Anthropic lists PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB and JSON, and says nothing about archives.
Gemini	Yes	Up to 10 files inside, up to 100 MB, no video or audio.

As of October 2026, from OpenAI's supported file types and its support team's forum reply, Anthropic's upload guide and Google's Gemini file limits.

Unzipping it yourself runs into the file limits.

OpenAI's suggested workaround is to extract the ZIP and upload the files one by one. We counted the files in 60 public code repositories, the kind of folder people zip for an AI, and the typical one held 183. Almost every one is over each vendor's file limit. One combined text file counts once everywhere.

Limit	Over it	Set by
10 files in a ZIP	60 of 60	Gemini
20 files in a chat	59 of 60	Claude
80 files every 3 hours	51 of 60	ChatGPT, unzipped by hand

Files in a fresh clone of each repository, counted September 2026 for our repository size study. A GitHub ZIP holds the same files.

From a ZIP to one file ChatGPT reads.

1
Drop the ZIP
Drop it in or pick it. .zip, .7z, .rar, .tar and .tar.gz are unpacked in your browser, folders inside included.
2
It reads everything inside
PDFs, Word, Excel and code become text in your browser, each under its own path, with the token count shown as it goes.
3
Upload the one file
Set Format to Plain and download the .txt. TXT is on both OpenAI's and Anthropic's lists of supported types, and there is no archive left to open.

Common questions.

Can I upload a ZIP file to ChatGPT?: Yes, the upload goes through. Reading it is a separate step: ChatGPT's file-analysis tools have to unpack the archive before it sees any file inside. ZIP is not on OpenAI's list of supported file types. OpenAI's support team wrote on its community forum in September 2026 that the file-analysis tools may not be able to unpack or inspect the archive, even when the upload appears to succeed.
Why does ChatGPT say it can't open my ZIP?: Because unpacking runs in ChatGPT's file-analysis tools, and when they error the archive stays closed. Forum reports from 2024 to September 2026 describe ZIPs that uploaded fine and then could not be read, including inside Projects and Business workspaces. A plain text file needs no unpacking, so the same contents combined into one file avoid the step that fails.
What does OpenAI recommend instead?: Its support reply suggests extracting the ZIP on your computer and uploading the individual files. Each of those files counts toward the upload quota: 80 files every 3 hours, and 3 a day on the Free plan, as of September 2026. In our sample of 60 public code repositories, 51 held more than 80 files. Combined into one text file, the whole archive spends one upload.
Can Claude read ZIP files?: ZIP is not on the list of document types Anthropic says Claude works with (PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB and JSON). Its help pages say nothing about archives. A chat also takes at most 20 files, and 59 of the 60 repositories we measured held more than that. One combined TXT file is on the list and takes one of the 20.
Can Gemini read ZIP files?: Yes, Google lists ZIP as supported, with limits: a ZIP can contain up to 10 files, can be up to 100 MB and can't include video or audio. Every one of the 60 repositories we measured held more than 10 files, the median 183. For code, Gemini's Import Code takes one folder or GitHub repository of up to 5,000 files instead. For documents, a combined text file carries everything in one.
Which archives does FileConcat open?: ZIP, TAR, TAR.GZ (or .tgz), GZ, 7Z, RAR, BZ2 and XZ, all unpacked in your browser. The documents inside, PDF, Word, Excel and PowerPoint included, are read as text the same way loose files are. A password-protected or damaged archive won't open; unpack that one on your computer and drop the folder.
Are my files uploaded to a server?: Not to us. FileConcat unpacks the archive in your browser, and there is no account to create. The only upload is the one combined file you choose to send to ChatGPT.

Hand ChatGPT what is inside the ZIP.

Not zipped yet? See how to upload a folder to ChatGPT. Each file still counts toward the ChatGPT file upload limit, and what each format inside keeps is on the formats list.