ChatGPT Not Working? 8 Common Errors and How to Fix Them

It always happens at the worst time. You're mid-prompt for a deadline, you hit send, and ChatGPT either gives you a red error message, freezes mid-response, or just stares back at you with nothing.

I've hit every one of these errors at least once. Most have a fast fix. Some require waiting it out. A few signal a real problem you need to address. Below are the 8 most common ChatGPT errors and what actually works to fix each one.

Quick checklist before you panic

Before you dive into specific error fixes, run these three checks. They solve about 60% of issues:

1. Check OpenAI's status page at status.openai.com — if there's an outage, it's not on your end and you can stop troubleshooting
2. Try a hard refresh with Ctrl+F5 (Windows) or Cmd+Shift+R (Mac)
3. Try ChatGPT in a different browser or incognito mode — this rules out extension or cache issues

If those three don't resolve it, find your specific error below.

1. "Something went wrong. If this issue persists, please contact us through our help center."

What it means: Generic ChatGPT error. Could be anything from a server hiccup to your specific prompt triggering a content filter.

The fix:

1. Wait 30 seconds and re-send. Half the time, it's a transient server issue.
2. Shorten your prompt. Very long prompts (2,000+ words) sometimes trigger this on browser sessions.
3. Remove special characters or unusual formatting. Pasting from PDFs sometimes brings invisible characters that confuse the parser.
4. Try a new chat. A specific chat thread can get into a stuck state. Starting fresh often works.

If none of that fixes it, the prompt may be tripping a content filter. Reword the request.

2. "Network error. Please check your connection and try again."

What it means: Either your internet dropped, or ChatGPT's servers timed out the connection.

The fix:

1. Verify your internet works by loading another site
2. Disconnect and reconnect your VPN if you're using one — VPN issues are the #1 cause of network errors with ChatGPT
3. Switch from Wi-Fi to mobile data (or vice versa) to test
4. Disable browser extensions that might be intercepting requests (ad blockers, privacy tools)
5. Try the desktop or mobile app instead of the browser

The VPN issue is worth highlighting — ChatGPT specifically flags certain VPN exit nodes that are commonly used for abuse. If your network errors started after enabling a VPN, that's almost certainly the cause.

3. "Too many requests in 1 hour. Try again later."

What it means: You've hit the rate limit for the model you're using. Free users hit this faster on GPT-5; paid users hit it on the highest-tier models too.

The fix:

1. Wait the time displayed. The cooldown is usually 1–3 hours.
2. Switch to a smaller model — the model picker at the top of a chat lets you swap to a model that hasn't hit the limit
3. Consider upgrading if you regularly hit free-tier limits — Plus or Pro give significantly higher caps

There's no trick to bypass this. The rate limit is enforced server-side per account.

4. "There was an error generating a response."

What it means: ChatGPT started generating, then something broke partway through. Often related to the response getting unexpectedly cut off or hitting a content issue.

The fix:

1. Hit the regenerate button on the failed response — works about 70% of the time
2. Add "continue" as a follow-up prompt if you got partial output
3. Break long prompts into smaller chunks — if you asked for a 10,000-word document, the model timed out. Ask for sections instead.
4. Disable Code Interpreter or Web Browsing if you don't need them — these features add complexity that occasionally fails

If this happens repeatedly on the same prompt, the prompt is the problem. Rephrase and try again.

5. ChatGPT is stuck on "thinking" forever

What it means: The response started but never came back. Could be a server issue, a network drop, or the model getting confused on a complex prompt.

The fix:

1. Wait 60 seconds. Long responses on complex prompts genuinely take that long sometimes.
2. Refresh the page — the response is often there when you reload
3. Check OpenAI's status page for partial outages
4. Stop the response and regenerate — there's a stop button next to the input field while it's thinking

If responses consistently hang for over a minute, OpenAI is probably having capacity issues. Wait it out.

6. "Conversation not found" or chats disappearing

What it means: Either OpenAI is having a sync issue, or you're signed into a different account than you think.

The fix:

1. Verify which account you're logged into — check the email in your profile menu. People with multiple accounts (work and personal) often think their chats are deleted when they've just switched accounts.
2. Refresh the page — temporary sync issues resolve on reload
3. Clear your browser cache for chat.openai.com (last resort, won't actually delete real chats)
4. Wait 24 hours — OpenAI sometimes has data recovery delays. Chats often reappear.

If a specific chat is genuinely gone after 48 hours, contact OpenAI support. Lost chats from server issues are eligible for partial credit.

7. ChatGPT is responding but the answers are wrong, generic, or hallucinated

What it means: Not technically an error, but a real problem. Could be a degraded model period, a poorly-formed prompt, or you hitting the smaller fallback model.

The fix:

1. Check which model is selected at the top of the chat. If you're on a smaller fallback model after hitting your limit, output quality drops.
2. Be more specific in your prompt. Vague prompts get vague answers. (See our prompts guide for techniques.)
3. Add a verification step — "Before responding, double-check your answer for factual errors."
4. Compare against another model — if Claude or Gemini gives a better answer to the same prompt, ChatGPT might be having a degraded period.

OpenAI occasionally A/B tests model variants, so output quality can vary slightly between users on the same model. This is normal but frustrating.

8. "You've reached the maximum length for this conversation."

What it means: Conversations have a context window limit. Very long chat threads eventually hit it.

The fix:

1. Start a new chat. Copy any context from the old one that the new one needs.
2. Use Projects (paid feature) to maintain shared context across many chats without hitting the limit per-chat
3. Use file uploads instead of long pastes — uploading a document is more efficient than pasting it inline

Long conversations also get slower and more error-prone before they fail entirely. As a rule, start a new chat after 30–40 message exchanges even if the old one still works.

Errors that signal you might be banned or restricted

Most ChatGPT errors are temporary. A few signal something more serious.

Repeated "We're not able to process your request" with no other detail: Could indicate your account has been flagged. Check your email for any messages from OpenAI. Try logging out and back in.

Account suspended messages: OpenAI rarely suspends accounts but does so for content policy violations. The email explaining the suspension will arrive separately.

Payment-related errors: "Your subscription is paused" or "We couldn't process your payment" — fix in the billing page. Common cause: expired credit card or country-specific payment issues.

When to switch tools temporarily

If ChatGPT has a major outage, you have options. The same prompts work on:

• Claude (claude.ai) — comparable quality, free tier available
• Gemini (gemini.google.com) — Google's chat AI, free tier available
• Perplexity (perplexity.ai) — search-focused but handles general prompts

Bookmark these as backup tabs. Outages happen on every platform; having an alternative ready means you don't lose an hour of work to a status page incident.

How to avoid most errors before they happen

A few habits that reduce error frequency:

1. Keep prompts under 1,500 words when possible. Very long prompts are more error-prone.
2. Don't paste raw PDF content — convert to plain text first, or upload the file directly instead
3. Use a stable internet connection for long sessions; mobile data over poor signal causes most network errors
4. Update your browser regularly — old browser versions have known issues with the ChatGPT interface

The takeaway

Most ChatGPT errors are temporary and fix themselves within minutes. The fastest path is usually:

1. Refresh the page
2. Try a new chat
3. Switch browsers or devices
4. Check OpenAI's status page

If those don't resolve it, the specific fixes above cover the vast majority of cases. Bookmark this post — you'll need it eventually, and the answer to "why isn't ChatGPT working" is almost always one of these eight things.