ChatGPT Not Working? Fix It in 5 Steps (Browser, App, API)
TL;DR Check status.openai.com first — most “ChatGPT not working” reports coincide with server-side incidents. Clear your browser cache or force-close the mobile app before anything else; stale sessions cause the…
- Check status.openai.com first — most “ChatGPT not working” reports coincide with server-side incidents.
- Clear your browser cache or force-close the mobile app before anything else; stale sessions cause the majority of client-side failures.
- If only specific features fail (voice, image generation, plugins), the issue is likely plan-gated or region-restricted, not a bug.
- Open Chatgpt in a private/incognito window to rule out a bad browser session.
- Disable extensions, VPN, or strict privacy tools temporarily, then retry the same action.
- Check the official status page or app notice before assuming your account is broken.
- Try the web app and mobile/desktop app separately to isolate the failing surface.
Overview
“ChatGPT not working” covers a wide range of failures: the page won’t load, responses stop mid-sentence, the app crashes, or the API returns errors. The fix depends entirely on where the problem sits — OpenAI’s servers, your browser, your network, or your account.
This guide separates those layers. Work through the quick fixes first. If none apply, the numbered steps below walk through systematic diagnosis from the cheapest check (is it even down?) to the most involved (network and DNS issues). Each step is independent — skip any that don’t match your symptoms.
chatgpt.com and openai.com domains.- Visit status.openai.com — if there’s an active incident, wait it out.
- Hard-refresh the page: Ctrl + Shift + R (Windows/Linux) or ⌘ + Shift + R (Mac).
- Try an incognito/private window — this bypasses extensions and cached state in one move.
- Switch browsers entirely. If ChatGPT works in Firefox but not Chrome, the problem is browser-specific.
- On mobile, force-close the ChatGPT app and reopen it. On iOS, swipe up from the app switcher; on Android, use “Force Stop” in app settings.
Symptoms
ChatGPT failures cluster into a few recognizable patterns:
Blank or white screen. The page loads but no chat interface appears. Usually caused by a cached JavaScript bundle that’s out of date, or a browser extension interfering with rendering.
Infinite loading spinner. The prompt submits but no response arrives. May indicate server overload, a dropped WebSocket connection, or a content policy filter that silently blocks the response.
“Something went wrong” error. A generic catch-all. Can mean anything from an expired session to a backend timeout. The lack of specificity is the point — OpenAI intentionally keeps these vague.
Response cuts off mid-sentence. The model hit its output token limit, or the connection dropped. This is distinct from the model refusing to answer, which produces a specific refusal message.
Login loop. You authenticate successfully but get redirected back to the login page. Almost always a cookie or third-party-cookie issue.
Why This Happens
ChatGPT failures fall into four buckets. Identifying which one you’re in saves time.
Server-side (OpenAI). High traffic, deployments, or incidents. You can’t fix these — only wait. OpenAI’s infrastructure handles hundreds of millions of conversations; intermittent degradation is normal, not exceptional.
Browser/client-side. Stale cache, incompatible extensions (especially ad blockers and privacy tools), outdated browser versions, or disabled JavaScript. These account for most “it works for everyone else but not me” reports.
Network. Corporate firewalls, VPNs, DNS filtering, or ISP-level blocking. ChatGPT uses WebSocket connections that some network appliances terminate unexpectedly. This is the most common culprit in corporate or school environments — and the one people check last.
Account/plan. Feature availability varies by plan. Some capabilities require specific subscription tiers. If a feature worked yesterday and doesn’t today, OpenAI may have changed what’s included.
## Check OpenAI Status and Your Account
Start here. Every other step is wasted effort if OpenAI’s servers are down.
Go to status.openai.com and check for active incidents. Pay attention to the specific component — “ChatGPT” and “API” are tracked separately. An API outage won’t affect the web app and vice versa.
Next, verify your account:
– Log out completely, then log back in. Expired sessions often manifest as weird UI glitches rather than a clean “session expired” message.
– Check your plan status at Settings in your ChatGPT account. If your subscription lapsed or payment failed, features may degrade silently — you might still access the interface but lose access to certain models or capabilities.
– Try a different conversation. If only one specific chat thread fails, it may be corrupted. Start a new chat and test there.
If the status page shows green and your account is fine, move to step 02.
## Fix Browser and Extension Issues
Browser problems cause the majority of client-side ChatGPT failures. Here’s how to isolate them systematically.
**Test in incognito mode first.** This disables all extensions and uses a clean cookie jar. If ChatGPT works in incognito, one of your extensions is the culprit.
Common offenders:
– **Ad blockers** (uBlock Origin, AdBlock Plus) — these may block OpenAI’s analytics or WebSocket endpoints, breaking the chat stream.
– **Privacy extensions** (Privacy Badger, Ghostery) — they can block third-party authentication cookies that ChatGPT needs for session management.
– **VPN browser extensions** — these route traffic through servers that OpenAI may rate-limit or block.
– **Grammarly and similar input modifiers** — they inject scripts into text areas and can conflict with ChatGPT’s input handling.
To find the specific extension: disable all extensions, confirm ChatGPT works, then re-enable them one at a time until it breaks.
**Clear site-specific data.** In Chrome: open DevTools (F12), go to Application → Storage → Clear site data. This is cleaner than clearing all browsing data because it targets only the current site.
**Check your browser version.** ChatGPT’s frontend uses modern JavaScript features. Browsers more than two major versions behind may fail silently. Update to the latest stable release.
## Troubleshoot Network and DNS
If ChatGPT fails on your entire network but works on cellular data, the problem is between your device and OpenAI’s servers.
**VPN interference.** Disconnect your VPN and try again. Some VPN exit nodes are flagged or rate-limited by OpenAI due to abuse from other users on the same IP. If you need the VPN for work, try switching to a different server location.
**Corporate or school networks.** Many organizations block or filter AI services at the firewall level. ChatGPT requires access to `chatgpt.com`, `openai.com`, and their CDN subdomains over HTTPS (port 443). It also uses WebSocket connections (`wss://`) for streaming responses — some corporate proxies terminate WebSocket upgrades. Talk to your network admin if you suspect this.
**DNS resolution.** Try switching to a public DNS resolver:
# Google DNS
8.8.8.8 / 8.8.4.4
# Cloudflare DNS
1.1.1.1 / 1.0.0.1
On Windows, flush the DNS cache with `ipconfig /flushdns`. On macOS, use `sudo dscacheutil -flushcache; sudo killall -HUP mDNSResponder`.
**ISP-level blocking.** Rare in most countries but real in some regions. If DNS changes don’t help, check whether your ISP blocks OpenAI domains. A tool like `traceroute chatgpt.com` or `nslookup chatgpt.com` can reveal where the connection fails.
## Fix Mobile App Issues (iOS and Android)
The ChatGPT mobile app has its own failure modes separate from the web version.
**Force-close and reopen.** This sounds obvious but mobile apps cache state aggressively. A stale cached state can cause blank screens or login loops. On iOS, swipe up from the app switcher. On Android, go to Settings → Apps → ChatGPT → Force Stop.
**Update the app.** OpenAI ships frequent updates that sometimes include breaking changes. An outdated app version may fail to communicate with updated backend APIs. Check the App Store or Google Play for pending updates.
**Check app permissions.** If specific features fail — voice input not working, unable to upload images — verify that the app has the required permissions (microphone, camera, photo library) in your device’s settings.
**Reinstall as a last resort.** If force-closing and updating don’t help, delete the app and reinstall it. This clears all local cache, stored credentials, and any corrupted data. You’ll need to log in again.
## Diagnose API-Specific Failures (Developers)
If you’re hitting the OpenAI API programmatically and getting errors, the diagnosis is different from the consumer web app.
**HTTP 429 — Rate limited.** You’re sending too many requests. Check your rate limits in the OpenAI platform dashboard. Implement exponential backoff. Don’t retry immediately — that makes it worse.
**HTTP 401 — Authentication failed.** Your API key is invalid, expired, or doesn’t have the required permissions. Regenerate the key in the dashboard if needed. Check that you’re passing it as `Authorization: Bearer sk-…` and not in a query parameter.
**HTTP 500 / 503 — Server error.** OpenAI’s side. Check the status page. If the status page says everything is operational but you’re getting 500s, the issue may be specific to your request — excessively long prompts, malformed JSON, or unsupported parameters for the model you’re calling.
**Timeout / connection reset.** Streaming responses (`stream: true`) use server-sent events. If your HTTP client has a short read timeout, it may drop the connection before the model finishes generating. Set read timeouts to at least 60 seconds for long completions.
# Example: setting timeout in the OpenAI Python SDK
from openai import OpenAI
client = OpenAI(timeout=120.0) # 120-second timeout
Most ChatGPT failures resolve at step 01 (it’s an OpenAI issue) or step 02 (it’s your browser). If you reach step 05, you’re likely dealing with an API integration problem, not a ChatGPT problem.
When Not to Troubleshoot
Some “ChatGPT not working” reports aren’t bugs — they’re features behaving as designed.
Content refusals. If ChatGPT refuses to answer a specific prompt but responds to others, it’s likely a content policy filter, not a malfunction. Rephrasing may help, but the filter is intentional.
Model differences. Different models have different capabilities. If a feature worked with one model but not another, check whether that model supports the feature you’re trying to use.
Slow responses during peak hours. ChatGPT can be noticeably slower during high-traffic periods (typically US business hours and evenings). Slowness isn’t the same as broken — if you’re getting responses, it’s working.
Region-based limitations. OpenAI’s services aren’t available in all countries. If you’re traveling or using a VPN that exits in an unsupported region, you may get blocked. This is by design, not a bug.
If It Still Fails
If none of the steps above resolve your issue:
- Capture the exact error. Screenshot the error message, note the browser/app version, and record the time. “It doesn’t work” is not actionable — exact details are.
- Check the OpenAI Community Forum for other reports of the same issue. If dozens of people report the same error simultaneously, it’s a server-side problem that OpenAI is likely already aware of.
- Contact OpenAI support. Use the help widget inside ChatGPT (click the
?icon in the bottom-right) or visit the help center. Include your error details, browser version, and what you’ve already tried. - Use an alternative interface temporarily. The ChatGPT API, third-party clients, or even Microsoft Copilot (which uses the same underlying models for some features) can serve as stopgaps while the primary interface is down.
Isolate the failing surface
Repeat the same chatgpt not working check in another browser, private window, or app so you know whether the issue is local or account-wide.
Remove local blockers
Temporarily disable extensions, VPN, custom DNS, or network filters, then refresh the session and try again.
Escalate only after the basics fail
If the issue follows your account across devices, collect the exact error text and contact the vendor support channel with that evidence.