Errors & Bugs

Troubleshoot error messages and unexpected behavior.

Explore this topic across different AI tools to compare symptoms, common root causes, and the fastest troubleshooting paths.

Browse Errors & Bugs by Tool

ChatGPT error 403 – How to Fix

A ChatGPT 403 error means access was forbidden — the most common causes are expired session tokens, browser extensions blocking requests, or VPN/proxy IP mismatches. Fix it by refreshing the page, clearing cookies for chat.openai.com, and retrying in incognito mode; if the error appears on long prompts, split them into smaller parts under 2,000 words.

ChatGPT Error in Message Stream: How to Fix It

The 'Error in message stream' in ChatGPT means the real-time streaming connection that delivers words to your screen was interrupted mid-response — not that your message was lost. This error appears in roughly 60–70% of cases due to an unstable network connection or a VPN endpoint throttling the persistent connection; switching to a hotspot and disabling VPN resolves it for most users.

How to report a ChatGPT bug effectively (what to include)?

To report a ChatGPT bug effectively, include the exact error text, a timestamp, your OS and browser/app version, and clear steps to reproduce — reports missing even one of these 4 elements are resolved 2–3x slower by support. Start with a clean session (sign out, clear cache/cookies, disable extensions) to confirm the bug is reproducible, then submit via help.openai.com.

How to fix ChatGPT messages not saving or missing in history?

ChatGPT conversation history is disabled by default in Temporary Chat mode and can also stop saving if Chat History is turned off in Settings — check Settings → Data Controls → Improve the model for everyone to confirm history is enabled. If history was on and messages are still missing, they are usually recoverable by refreshing on the same account and browser.

ChatGPT Network Error: Why It Happens and How to Fix It

A ChatGPT network error means the connection between your browser and OpenAI's servers was interrupted. The most common causes are unstable internet, VPN interference, browser extension conflicts, or OpenAI server issues. Fix it by refreshing, switching networks, disabling VPN, or trying incognito mode. If the error only appears on long responses, enable the regenerate option and break your prompt into shorter parts.

ChatGPT Something Went Wrong Error: Fix Guide

The 'Something went wrong' error in ChatGPT is a generic catch-all that appears for dozens of different underlying causes — session expiry, browser extension conflicts, network throttling, or temporary server issues. The fastest fix for 70% of cases is signing out, clearing cookies for openai.com, and signing back in from a private browser window.

ChatGPT Something Went Wrong: 7 Fixes That Actually Work

The 'Something went wrong' error in ChatGPT is a generic catch-all that usually means a session problem, a corrupted browser cache, or a temporary server issue. Fix it by signing out and back in, clearing your cache, or trying a private browser window. If it appears on every conversation, your account session may be corrupted and needs a full reset.

ChatGPT Voice Not Working: Fix Microphone Permissions, Browser Settings, and Audio Issues

ChatGPT voice input requires microphone permission granted at both the browser level and the OS level — if either is blocked, voice will fail silently or show a microphone error. Check browser permissions first (the padlock icon next to the URL shows mic status), then check OS-level microphone privacy settings, which on macOS and Windows can block permissions even when the browser appears to have them enabled.

Claude Attachments Not Uploading or Failing to Process? Here's the Fix

Claude Pro supports file uploads up to 10 MB per file; supported formats include PDF, Word (.docx), plain text, images (PNG/JPG/GIF/WebP), and most code files. If your attachment fails or won't process, the most common causes are an oversized file, an unsupported format, a stale browser session, or a network/extension block. Clear site data for claude.ai, disable ad blockers, and retry in incognito — that resolves the issue in the majority of cases.

Claude Blank Page / White Screen: Fix Cache, Extensions, and Script Blocking

A Claude blank page or white screen is almost always caused by a stale browser cache, a script-blocking extension, or a VPN blocking WebSocket connections. Open an incognito window and sign in — if Claude loads there, clear all site data for claude.ai in your main browser and disable extensions one at a time. If Claude is blank in incognito too, disable your VPN and check status.anthropic.com for active outages.

Claude error loading chat – How to Fix

Claude's 'error loading chat' message appears when a conversation fails to sync from the server or a browser extension blocks the loading request. Refresh the page once, then try opening the conversation in an incognito window. If a specific chat refuses to load on all devices, that conversation's data may be corrupted on Anthropic's servers — start a new conversation to continue your work.

How to report a Claude bug effectively (what to include)?

Report Claude bugs at support.anthropic.com. An effective bug report includes: the exact error message, the specific steps that reproduce it (not just 'it stopped working'), your browser and OS version, whether it happens in incognito, and whether it affects all conversations or just one. Including a screenshot and the approximate timestamp speeds up resolution significantly.

Claude Something Went Wrong: Causes and Fixes

The 'something went wrong' error in Claude is a generic message that covers session issues, browser problems, and temporary server errors. Start by refreshing the page and signing out then back in. If the error persists, clear your browser cookies for claude.ai and try a private window. This error almost never indicates an account problem — it is usually a temporary connection or session issue.

How to fix Claude error while generating (stops mid-response)?

Claude stopping mid-response is usually caused by a network interruption, a browser extension blocking the streaming connection, or the response hitting a length or usage limit. The fastest fix is to refresh the page, disable extensions, and retry with a more focused request. If Claude consistently stops at the same point in a response, the content itself may be triggering a safety check — try rephrasing the request.

GitHub Copilot Not Working: Fixes for VS Code, JetBrains, and CLI

GitHub Copilot stops working most often because of an expired or inactive subscription, a failed authentication token, a firewall blocking GitHub's AI endpoints, or a version mismatch between the Copilot extension and your editor. Start by checking your subscription at github.com/settings/copilot, then sign out and back into the extension. Most issues resolve with a token refresh or extension update.

Cursor Not Working: Fixes for the Most Common Errors

When Cursor stops working, the most common causes are a failed AI model connection, an expired API key or subscription, a corrupted extension state, or a VS Code compatibility issue. Start by checking your Cursor account status at cursor.com, then reload the window with Ctrl+Shift+P → Reload Window. Most Cursor issues are resolved by signing out and back in or reinstalling the app.

Gemini Not Working: 8 Fixes for the Most Common Errors in 2026

When Gemini stops working, start with the basics: check Google's status page, clear your browser cache, sign out and back in with your Google account, and try a different browser. Most Gemini outages and loading errors resolve within minutes using one of these steps. If Gemini Advanced features are missing, check that your Google One subscription is active.

How to fix Claude errors?

Start by checking whether the issue is caused by account access, plan status, browser state, or a temporary service incident. Then follow the step-by-step checks below to isolate the root cause quickly.

How to fix OpenClaw error?

Start by checking whether the issue is caused by account access, plan status, browser state, or a temporary service incident. Then follow the step-by-step checks below to isolate the root cause quickly.

Midjourney creation failed internal server error – How to Fix

Midjourney's 'creation failed: internal server error' means the job failed on Midjourney's GPU servers — this error occurs on the server side, not your device. Wait 2 to 5 minutes and retry your prompt, check status.midjourney.com for outages, and simplify your prompt if failures persist. Most internal server errors resolve within a single retry.

Midjourney error failed to download image – How to Fix

The 'failed to download image' error in Midjourney occurs when the generated image cannot be delivered from Midjourney's CDN to your client — refresh the page and click the image again, try right-clicking and selecting 'Open image in new tab,' or use the web gallery at midjourney.com to access the generated image directly.

Midjourney error loading image – How to Fix

Midjourney's 'error loading image' typically means a reference image URL is invalid, blocked, or too large — use a direct image URL (ending in .jpg, .png, or .webp) rather than a webpage URL, ensure the image is publicly accessible without login, and keep uploaded reference images under 25MB. If the error appears on a generated image, refresh and try again.

Midjourney error your unique trace – How to Fix

The 'your unique trace' error in Midjourney is a job tracking identifier included in error messages to help support diagnose failures — copy the trace ID from the error message, note the time it occurred, then retry your prompt. If the error repeats, submit the trace ID to Midjourney support at docs.midjourney.com so they can investigate the server-side failure.

How to report a Midjourney bug effectively (what to include)?

Midjourney bug reports submitted through their official channels (Discord #bug-reports channel or docs.midjourney.com support) are most effective when they include: the exact error message or behavior, the full prompt used, your plan tier (Basic/Standard/Pro/Mega), the specific channel or interface (Discord vs. web), timestamps, and screenshots or screen recordings. Reports without reproduction steps are rarely actionable.

How to fix OpenClaw 401/invalid API key errors for cron jobs?

A 401 error in OpenClaw cron jobs means the Anthropic API key stored in your agent's environment is invalid, expired, or missing — regenerate it from console.anthropic.com, update the ANTHROPIC_API_KEY environment variable in your OpenClaw workflow, and re-run the job. This error does not indicate an OpenClaw account problem; it is an Anthropic API authentication failure passed through to your cron output.

How to report a OpenClaw bug effectively (what to include)?

A well-structured OpenClaw bug report gets resolved 3 to 5 times faster than a vague one — include the exact error message, the run ID from the dashboard URL, your agent configuration YAML (with API keys removed), the sequence of steps that reproduces the issue, and your OpenClaw plan tier. Submit via the support form at openclaw.com or paste it into the #bug-reports channel on their Discord.

How to fix OpenClaw git push failing in automation (auth/DNS/network)?

Git push failures in OpenClaw automations are caused by one of three things: missing or expired credentials (no SSH key or PAT configured in the worker environment), DNS resolution failures in OpenClaw's container network that block GitHub/GitLab domains, or a repository permission issue where the token lacks write access. The fix in 90% of cases is adding a valid Personal Access Token with repo write scope to your OpenClaw workflow environment as GIT_TOKEN or configuring SSH key authentication.

Perplexity Citations Not Loading or Sources Missing – How to Fix

Perplexity citations fail to load in over 70% of cases because an ad blocker or privacy extension is blocking the source-fetching requests that run alongside the AI response. Disable your ad blocker (uBlock Origin, AdGuard, Ghostery) for perplexity.ai, reload the page, and run your query again — citations should appear as numbered blue links below the answer. If the problem continues, switch from Pro Search to Standard Search as a quick test, then clear your browser cache.

Perplexity Error 401 – How to Fix Unauthorized Error

Perplexity error 401 means Unauthorized — your session has expired or your API key is invalid. For web users, log out at perplexity.ai/settings/account, clear cookies, and log back in. For API users, check that your key is active at perplexity.ai/settings/api and that you are sending it as a Bearer token in the Authorization header. A 401 error is never caused by Perplexity's servers — it is always an authentication issue on your end.

Perplexity Error 403 – How to Fix Forbidden Error

Perplexity error 403 means Forbidden — your request was understood but blocked. The three most common causes are: regional restrictions (Perplexity is unavailable in some countries), account-level restrictions (your account was flagged or your subscription lapsed), and content policy blocks (the specific query was rejected). Start by disconnecting any VPN, then checking your account status at perplexity.ai/settings/account, then clearing cookies and logging back in.

Perplexity Error 500 – How to Fix Internal Server Error

Perplexity error 500 is an Internal Server Error — the problem is entirely on Perplexity's servers, not your device or network. There is no local fix. Check perplexity.ai/status to see if an incident is active, then wait 5 to 15 minutes before retrying. Most Perplexity 500 errors resolve automatically within 15 minutes as the server recovers. If the error persists beyond 30 minutes, report it at perplexity.ai/contact.

Perplexity Error in Processing Query: Causes and Fixes

The 'Error in Processing Query' message on Perplexity appears when your query is too long or complex for the model to process, contains phrasing that triggers content filters, or when a backend service times out mid-response. The fastest fix is to shorten your query to under 500 characters, remove ambiguous or sensitive phrasing, and resubmit. If the error persists across multiple queries, check perplexity.ai/status for an ongoing service incident.

Perplexity Error Messages – What They Mean and How to Fix

Perplexity error messages fall into three groups: rate limit errors (too many requests — wait 60 seconds or upgrade your plan), service errors (something went wrong, network error — refresh the page or check perplexity.ai/status), and query errors (error processing query, content policy — rephrase or shorten your search). The fix depends on which error you see. This guide covers the 10 most common Perplexity error messages with specific solutions for each.

Perplexity 'Something Went Wrong': API Failures and Service Timeouts Fixed

The 'Something Went Wrong' error on Perplexity signals a technical failure in the backend — typically an API call failure, model endpoint timeout, or web retrieval service error — rather than a user-facing input problem. This error is distinct from the generic 'Error in Processing Query' message, which usually involves query content. For 'Something Went Wrong,' wait 30 to 60 seconds and retry; if the error persists for more than 5 minutes, check perplexity.ai/status for an active incident.

Perplexity Error Today: Is Perplexity Down? How to Check and Fix

To check whether Perplexity is down right now, go directly to perplexity.ai/status — this is the official status page maintained by Perplexity's engineering team and shows live incident reports. If the status page shows all systems operational but you are still experiencing errors, the problem is likely specific to your account, browser, or network rather than a platform-wide outage.

Perplexity Error – How to Fix Any Error on Perplexity

Most Perplexity errors resolve in under 5 minutes by following this order: identify the error type (401/403/500/network), refresh the page once, clear browser cache and cookies, switch to a different browser or network, then check perplexity.ai/status for any active outage. If the status page shows no incident, the error is almost always local — a cache or network fix will resolve it.

How to report a Perplexity bug effectively (what to include)

To report a Perplexity bug effectively, email support@perplexity.ai with 6 pieces of information: the exact error message text, the time and date of the issue, your browser and OS version, the steps needed to reproduce the bug, whether it occurs in an incognito window, and a screenshot. Perplexity's support team typically responds within 24 to 48 hours on business days. For widespread outages, also tag @perplexity_ai on X for faster acknowledgment.

Perplexity File Analysis Failing or Processing Error – How to Fix

Perplexity file analysis fails most often because the uploaded file is password-protected, corrupted, or exceeds the size limit — PDF files must be under 25 MB and must not have DRM or password protection. Remove the password using your PDF viewer, re-export the file at a smaller size, and upload again. If the error persists on a valid file, clear your browser cache and try uploading from an incognito window.

Perplexity Network Error – How to Fix

A Perplexity network error means your connection to Perplexity's servers is being interrupted — the problem is between your device and Perplexity, not on Perplexity's servers. Fix it in this order: disconnect your VPN completely, switch your DNS to 8.8.8.8 or 1.1.1.1, clear your browser cache, then switch to a different network such as your phone's mobile hotspot. If Perplexity works on mobile data but not your main connection, the issue is your router, ISP, or corporate firewall.

Perplexity not working – How to Fix

Perplexity goes down for 1–4 hours a few times per month; if the site loads but searches fail, the fastest fix is opening an incognito tab, disabling browser extensions, and switching to a different network. If that does not help, check status.perplexity.ai and wait for the incident to resolve — forced retries will not speed up recovery.

Perplexity Search Not Loading or Blank Results – How to Fix

When Perplexity shows a blank page or the search results area stays empty after submitting a query, the cause is almost always a JavaScript rendering failure — not a problem with Perplexity finding results. The fix in 80% of cases is disabling your ad blocker for perplexity.ai and doing a hard reload (Ctrl+Shift+R or Cmd+Shift+R). If that does not work, clearing your browser cache and testing in an incognito window will isolate whether an extension or corrupted cached file is preventing the page from rendering.

Perplexity Search Not Returning Results – How to Fix

When Perplexity returns no results, the most common causes are a network connectivity issue, an overly long or complex query that times out, or a temporary service disruption. In most cases, shortening your query to under 100 words and refreshing the page resolves it within 30 seconds. If results are consistently empty, check perplexity.ai/status for active incidents and try switching from Pro Search to Standard search mode.

Perplexity Something Went Wrong + Network Errors – How to Fix

When Perplexity shows 'Something went wrong' alongside a network error, the most likely cause is a DNS resolution failure, a VPN or proxy interrupting the connection, or a corporate firewall blocking Perplexity's API endpoints. Disconnect your VPN, switch to a public DNS server (8.8.8.8 for Google or 1.1.1.1 for Cloudflare), and reload the page. If the error appears on a work or school network, test via mobile hotspot — if it works on hotspot, the institution's firewall is the cause.

Perplexity Something Went Wrong – How to Fix

The 'Something went wrong' error on Perplexity is a generic server-side error that resolves on its own in over 80% of cases with a simple page refresh. If refreshing does not help, the three most common root causes are: a temporary service disruption (check perplexity.ai/status), a VPN or network interference blocking the response stream, or an expired login session. Reload the page, clear your browser cache, and log out then back in — most users are back to normal within 2 minutes.

Perplexity Too Many Requests (429): How to Fix It

A Perplexity 'Too Many Requests' or 429 error means either you sent requests too fast (API: wait 60 seconds, then retry with backoff) or your weekly Pro search quota of 200 searches is exhausted (wait until Monday 00:00 UTC or switch to Standard search). Temporary burst limiting clears in under 2 minutes; quota exhaustion lasts until the weekly reset.