Claude 4 API Error Codes Complete Reference

Change the model name in your code:

# Python — change one line
response = client.messages.create(
    model="claude-opus-4-8",  # was "claude-4-opus"
    messages=[{"role": "user", "content": "Hello"}]
)

# Node.js — change one line
const response = await client.messages.create({
  model: "claude-opus-4-8",  // was "claude-4-opus"
  messages: [{ role: "user", content: "Hello" }]
});

Same as 410 — update the model name. Also update your SDK to the latest version:

# Python
pip install --upgrade anthropic

# Node.js
npm update @anthropic-ai/sdk

Check the max output tokens for your new model:

1. Truncate your input or summarize long documents
2. Use max_tokens to limit output, freeing input space
3. Switch to a model with a larger context window if needed

1. Check ANTHROPIC_API_KEY environment variable is set
2. Ensure you're using the x-api-key header (not Authorization: Bearer)
3. Generate a new key from console.anthropic.com

# Correct header format
curl https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{"model": "claude-opus-4-8", "messages": [...]}'

# ❌ WRONG — this causes 401
curl https://api.anthropic.com/v1/messages \
  -H "Authorization: Bearer $ANTHROPIC_API_KEY" \

1. Check your Anthropic plan — some models require higher tiers
2. Verify organization permissions in the Anthropic Console
3. Contact Anthropic support if you believe you should have access

Verify your base URL:

# Correct base URL
https://api.anthropic.com/v1/messages

# ❌ Common mistakes
https://api.anthropic.com/v1/chat/completions  # OpenAI format
https://api.anthropic.com/messages  # Missing /v1

1. Add retry logic with exponential backoff
2. Check your rate limits in the Anthropic Console
3. Request a rate limit increase if needed

# Python — retry with backoff
import time

def call_with_retry(fn, max_retries=3):
    for i in range(max_retries):
        try:
            return fn()
        except anthropic.RateLimitError:
            if i == max_retries - 1:
                raise
            time.sleep(2 ** i)  # 1s, 2s, 4s

1. Retry after 30-60 seconds
2. Check status.anthropic.com for incidents
3. If persistent, switch to a fallback model

1. Retry with exponential backoff (start at 5 seconds)
2. Implement a fallback model (e.g., GPT-5 or Gemini)
3. Avoid peak migration hours if possible

1. Increase your timeout (default is 600s in Python, 60s in Node)
2. Reduce input length
3. Use streaming for long outputs

# Python — increase timeout
client = anthropic.Anthropic(timeout=300.0)  # 5 minutes

# Node.js — increase timeout
const client = new Anthropic({ timeout: 300000 });  # 5 minutes

1. Implement stream reconnection logic
2. Handle partial responses gracefully
3. Use max_retries parameter for automatic retries

# Python — enable auto-retries
client = anthropic.Anthropic(max_retries=3)

# Node.js — enable auto-retries
const client = new Anthropic({ maxRetries: 3 });