ChatGPT Troubleshooting: Fixing “Message Failed to Send” and 400 Request Errors

If you encounter ChatGPT “Message failed to send” or a 400 request error, in most cases it’s not that your account is broken—it’s that your network, browser cache, or the content you entered triggered a block. Below is a troubleshooting checklist ordered from fastest to slowest. You can usually identify the cause and restore the conversation within 10 minutes.

Start with two quick checks: service status and the network path

First, open OpenAI’s status page to confirm whether there is a widespread outage; if the server is unstable, refreshing will often lead to repeated errors. Second, troubleshoot the network: temporarily disable your proxy/VPN, corporate gateway, or ad-blocking DNS, then try again using a mobile hotspot to determine whether the path is being blocked.

If it works immediately after switching networks, the issue is basically in your local network policy or outbound IP. Next, prioritize checking router/proxy rules instead of messing with the account.

Browser-side troubleshooting: cache, extensions, and login state

“Failed to send” is most commonly caused by the browser environment: first, log in once in an incognito window. If it works, it indicates your cache or extensions are interfering. Continue troubleshooting by clearing site data (cookies and cache) for chat.openai.com, and temporarily disabling script-related extensions (ad blockers, Tampermonkey, privacy/anti-tracking tools).

Also check whether frequent account switching across multiple devices is causing a login-state conflict: sign out of all devices and sign in again to avoid cases where you appear logged in but requests are not carrying a valid session.

Common triggers for 400 request errors: oversized content and special input

A 400 is usually because the request content is judged to have an “invalid format” or to be “too large.” When troubleshooting, first shorten what you input: send long paragraphs in batches; reduce the number of images/files or use smaller ones, then try uploading or sending again.

Another subtle cause is abnormal characters introduced by copy-paste (for example, control characters pasted from tables or PDFs). The most reliable approach is to paste the content into a plain-text editor first, then copy it back into ChatGPT, or manually type a short sentence to test whether it can be sent normally.

If it’s still unstable: recovery steps—switch chats, switch entry points, and submit details

If the same conversation keeps erroring, troubleshooting suggests creating a new chat to test; some conversations are more likely to trigger exceptions when the context becomes too heavy. You can also cross-check by using another browser or the official mobile app to quickly rule out a single-client issue.

If it still fails, record the error message, the steps to reproduce, the network environment (whether a proxy is used), and the approximate input length, and submit feedback along with screenshots. The more complete the information, the easier it is to identify a specific block or compatibility issue and avoid repeated trial and error.