Claude AI Not Working — Diagnostic Checklist

If Claude AI is not working, first check status.claude.com; if the status page looks normal, the problem is usually a session issue, browser problem, usage cap, or model-specific limit rather than a full outage. As an independent guide to Claude by Anthropic, c-ai.chat tracks the common causes below and links up to our main Claude guide for broader context.

• Is Claude down right now?
• Common failure modes (not actually outages)
• Recent incident history
• What to do during an outage
• Other questions readers ask
• The honest take

• Check the official status page first before changing settings
• Many “Claude down” reports are actually browser, session, or usage-limit issues

Is Claude down right now?

The fastest way to check live status is direct: check status.claude.com. That page is the canonical source for current incidents affecting Claude chat, the API, authentication, and related services.

If the status page shows degraded performance, partial outage, or a major incident, the issue is probably not on your side. If it shows all systems operational, you should move to the checklist in the next section. You can also compare what you see in the web app with the developer side of Claude on our API guide, because some problems affect chat access but not API calls, or the reverse.

A quick way to narrow it down:

1. Check the official status page

   Open status.claude.com. If there is an active incident, wait for mitigation notes before troubleshooting further.

2. Test another surface

   Try Claude on a different browser, device, or network. If one surface fails and another works, it is likely local rather than platform-wide.

3. Identify whether it is web or API only

   If you use the API, compare results in platform.claude.com with the chat app on claude.ai or claude.com.

4. Look for the exact error pattern

   Login loop, blank screen, slow responses, and “message failed” errors usually point to different fixes.

Common failure modes (not actually outages)

A lot of “Claude is down” searches come from user-side problems that look like outages. These are the most common ones, along with the fix that usually works.

• Rate-limit hits or usage caps. On the free plan, usage limits can block new messages or reduce availability during busy periods. On paid plans, especially long chats or heavy file work can hit practical limits. Fix: wait for the limit window to reset, start a shorter thread, reduce attachments, or move to a lower-cost model or API workflow if you need steadier throughput.
• Region or network issues. A VPN, corporate firewall, hotel Wi-Fi, or local ISP problem can break login, loading, or message delivery while the service itself is fine. Fix: turn off the VPN, switch networks, try mobile data, or test from another location.
• Browser cache, extension, or cookie problems. Privacy extensions, script blockers, stale cookies, and aggressive browser settings can cause blank pages, stuck spinners, or sign-in loops. Fix: open Claude in a private window, disable extensions, clear site data for claude.ai and claude.com, then sign in again.
• Expired sessions. Claude may appear frozen or fail silently if your auth session has expired in the background. Fix: sign out, refresh, and sign back in. If that fails, clear cookies for the Claude domains and retry.

There are also model-specific cases. If one conversation using a larger model stalls while another short chat works, that points to a capacity or context issue rather than a global outage. Claude supports multiple models with different cost and performance trade-offs, and some problems are easier to route around by switching models or simplifying the request. Our Claude features guide covers where the main app and model capabilities differ.

Symptom	Likely cause	Best first fix
Blank page or endless spinner	Browser cache, extension conflict, site data issue	Use private mode, disable extensions, clear cookies
Login loop	Expired session or blocked auth flow	Sign out, clear site data, retry on another browser
Messages fail to send	Usage cap, temporary capacity issue, unstable network	Wait, shorten prompt, switch network, retry later
One model slow, others normal	Model-specific load or long context strain	Use a smaller model or shorter context
Claude works on phone but not desktop	Desktop browser or extension issue	Test clean browser profile or different browser
API works but chat does not	Web app issue, auth problem, or front-end outage	Check status page and continue via API if urgent

Recent incident history

In recent weeks and recent quarters, the issues users most often report have been temporary chat unavailability, elevated error rates, slower responses, authentication trouble, and occasional API disruption during periods of high load or maintenance. That pattern is normal for a large AI service, and the official record is the one that matters: use status.claude.com for canonical incident logs, while trust.anthropic.com covers security and trust information rather than live uptime.

What to do during an outage

If Claude is having a real outage, the goal is not endless refreshing. It is to keep moving with the least friction.

• Try the API instead of the web app. If chat access is failing but your work is time-sensitive, test whether platform.claude.com or your existing API workflow still works. Sometimes the issue is front-end specific.
• Use a smaller model. If one model is overloaded or timing out, try a lighter option. Claude Haiku 4.5 is the fast, cheap model at $1 per million input tokens and $5 per million output tokens; Claude Sonnet 5 is the usual default at $2/$10 (introductory through 31 August 2026; standard $3/$15 from 1 September 2026).
• Batch non-urgent work later. If your workflow can wait, postpone large jobs until service stabilises. Anthropic’s Batch API can also reduce cost by 50% on both input and output when you are not working interactively.
• Shorten prompts and reduce files. Long context and large attachments make failures more likely when systems are under load. Ask for one task at a time and split uploads if possible.
• Switch tools for the moment. If the deadline matters more than the model, use ChatGPT or Gemini temporarily, then move back once Claude recovers. That is a practical fallback, not a statement that one service is always better.

For developers, one useful rule is to separate “service unavailable” from “too much work in one request.” If a short prompt succeeds and a huge prompt fails, you are probably looking at request size, context management, or temporary capacity constraints rather than a complete outage. Our Claude FAQ covers several edge cases around limits, while the main Claude guide explains the official product surfaces and where each one fits.

Other questions readers ask

These are the related questions that usually sit next to “claude ai not working” in search results.

The honest take

If you searched for “claude ai not working,” the most honest answer is this: sometimes Claude really is having an outage, but a large share of failures come from local browser problems, expired sessions, usage limits, or one overloaded conversation. The official status page should always be your first stop, not social chatter and not third-party guesswork.

When the status page is clear, most users can get back in by switching browser, clearing site data, logging in again, changing network, or starting a shorter chat. When the status page shows an incident, stop troubleshooting, use a fallback if the task is urgent, and wait for the official update from Anthropic.

Independent guide. Not affiliated with Anthropic. For the official Claude product, visit claude.ai.

Last updated: July 16, 2026