DEV Community

brian austin
brian austin

Posted on

Claude.ai is down right now. Here's how to keep working without it.

#claude #ai #discuss #devops

Claude.ai is down right now. Here's how to keep working without it.

If you're reading this because Claude just returned a 503 or a vague 'elevated errors' message — you're not alone.

As of today, Claude.ai is showing a major outage with elevated API errors affecting both the consumer product and the API. Anthropic's status page confirms it.

Here's what to do right now.

Option 1: Switch to a Claude API wrapper that's still up

Some Claude API resellers and wrappers have separate infrastructure from Claude.ai's consumer product. If your primary issue is the chat UI being down — not the underlying API — this is your fastest fix.

SimplyLouie runs its own layer on top of the Claude API and has been online through today's outage. $2/month, no usage limits within fair use. 7-day free trial, card required but not charged for 7 days.

Option 2: Fall back to the raw Anthropic API directly

If you're a developer and the API itself is throwing errors, you need a fallback in your code. Here's a pattern:

async function claudeWithFallback(prompt) {
  const providers = [
    () => callAnthropic(prompt),
    () => callOpenAI(prompt),   // or any other provider
  ];

  for (const provider of providers) {
    try {
      return await provider();
    } catch (err) {
      console.warn('Provider failed, trying next:', err.message);
    }
  }
  throw new Error('All providers failed');
}
Enter fullscreen mode Exit fullscreen mode

This pattern is called a provider waterfall. Add it to any AI integration you care about.

Option 3: Check if it's just the UI, not the API

Claude.ai (the consumer chat product) and the Claude API are different services. Sometimes one is down while the other isn't.

Test the API directly:

curl https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-3-5-haiku-20241022",
    "max_tokens": 50,
    "messages": [{"role": "user", "content": "ping"}]
  }'
Enter fullscreen mode Exit fullscreen mode

If that returns a response, your code is fine — it's just the UI.

If you get a 529 (overloaded) or 500, the API itself is having issues.

Option 4: Local model as emergency fallback

For developers who need zero-dependency uptime:

# Install ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a capable model
ollama pull llama3.2

# Run it
ollama run llama3.2
Enter fullscreen mode Exit fullscreen mode

Llama 3.2 is not Claude. But for most tasks — summarization, code review, drafting — it's serviceable as an emergency fallback.

The real lesson from every Claude outage

Every time Claude goes down, the same thing happens: developers realize they've built a single point of failure into their workflow.

The solution isn't to switch providers permanently — it's to design for resilience:

  1. Don't depend on one AI provider for anything production-critical
  2. Cache responses for repeat queries where possible
  3. Have a fallback — even a dumber, local one
  4. Keep your prompts portable — don't write prompts that only work with one model's quirks

AI services go down. They'll keep going down. The providers building trillion-dollar infrastructure still have outages. That's fine — it's software.

What's not fine is having no plan when they do.

While you wait

The HN thread on this outage is active: Claude.ai unavailable and elevated errors. Real-time status from Anthropic: status.claude.com.

And if you want a $2/month backup that runs on Claude and stays up independently of Claude.ai's consumer infrastructure — simplylouie.com. 7-day free trial. No charge for 7 days.

Hope the outage resolves quickly. In the meantime — build for resilience.

Top comments (0)