“An unknown error occurred” in ChatGPT means a request failed without a specific reason. The answer never arrives, a regenerate fails, or a file you attached stops uploading. Treat it the way OpenAI treats a failed response: retry first, then rule out an outage, your browser and your network.

OpenAI’s troubleshooting guide for ChatGPT error messages says a failed response “may be a one-time error” and lists the steps below. They apply here because the message describes the same thing: ChatGPT tried to do something and the request did not complete.

On this page

Fix it in this order

  1. Click Regenerate or resend the message. OpenAI’s first step for a failed response is to regenerate, since many failures are one-off.
  2. Check for an outage. Open status.openai.com. If an incident is listed, wait for it to clear.
  3. Restart the browser or the app. Close ChatGPT completely and open it again. On the web, a hard refresh (Cmd/Ctrl + Shift + R) reloads the page without cached files.
  4. Turn off VPNs and proxies. OpenAI lists “VPNs or proxy connections” and “Web Protect or other security filters” among the causes of failed responses.
  5. Try a different browser or an incognito window. Incognito mode runs without most extensions and without your saved site data, which rules out a broken extension or stale cookies.
  6. Switch network or device. If ChatGPT works on your phone’s mobile data but fails on your home or office Wi-Fi, the network is the problem.
  7. Start a new chat. For replies that stall or fail inside a long conversation, OpenAI recommends a new chat “if the conversation is long or has many turns”.

If it happens while uploading a file

The same message can appear when an attachment fails to upload. Before you retry, check the file against the limits in OpenAI’s File Uploads FAQ:

  • every file has a hard limit of 512 MB;
  • text and document files are capped at 2 million tokens per file;
  • CSV files and spreadsheets cannot exceed roughly 50 MB, depending on row size;
  • images are limited to 20 MB each;
  • you can upload up to 80 files every 3 hours, and Free users are limited to 3 file uploads per day, with lower limits possible at peak times.

OpenAI also notes that failed upload attempts “can sometimes count toward the upload-rate cap”, so repeated retries of the same failing file can use up your allowance. If the file is within the limits, check status.openai.com for an upload incident, then try incognito mode or another browser before uploading again. On a company network, OpenAI’s network recommendations give the example of an upload that fails because the IT team needs to allowlist *.oaiusercontent.com.

Why the request fails

OpenAI groups the causes of ChatGPT errors into “network configuration, browser extensions, or transient server-side problems”. For this message that usually means one of:

  • A short problem on OpenAI’s side. These clear on their own, which is why Regenerate is the first step.
  • A connection that drops or gets filtered. VPNs, proxies, secure DNS tools and company firewalls can block or cut off the traffic ChatGPT needs.
  • The browser. Script blockers, privacy extensions and old cached data can break requests the page sends.
  • A long conversation. Long threads take longer to process and are more likely to stall.

When to contact OpenAI Support

If the error keeps coming back across browsers, devices and networks and the status page is clear, OpenAI asks you to collect a HAR file and the browser console errors with timestamps, note the model and the conversation URL, and send them to Support. For upload problems, include your account email, a screenshot, the time and timezone, the browser or app, and the request ID if you have one. Support is reachable from the chat widget at help.openai.com.