Status card
Provider status guide
Is Replicate down?
Check official Replicate status and troubleshoot common API errors before blaming your code, key, quota, model name, network, or the provider.
Common errors
What usually breaks
401
Invalid or missing API key
Authentication failed before the request reached the model.
403
Permission denied
The key is recognized but not allowed to use this model, endpoint, region, or project.
429
Rate limit or quota exceeded
Your request rate, token throughput, or quota exceeded the provider's limit.
500
Provider server error
The provider accepted the request path but failed internally.
503
Unavailable
The service or model is temporarily unavailable.
timeout
Timeout
No response was available before your client timeout expired.
Troubleshooting checklist
Before calling it a Replicate outage
- Check the official provider status page.
- Confirm your API key is valid and belongs to the right project.
- Confirm your account has credits, quota, and model access.
- Confirm the model name and endpoint path are correct.
- Retry with exponential backoff instead of immediate loops.
- Test with cURL outside your app.
- Try another model or provider if production is impacted.
- Enable fallback routing before the next incident.
Decision tree
Official status is green, but my Replicate API still fails
- Run a minimal cURL command with the same key and model outside your app.
- If cURL fails with 401 or 403, check key scope, project, account access, and model permissions.
- If cURL returns 429, check quota, billing, token-per-minute limits, and concurrency.
- If only browser requests fail, treat it as CORS or preflight until cURL proves otherwise.
- If one model fails but another works, check model availability or fallback routing.
- If production fails but local works, compare environment variables, outbound network rules, and region.
FAQ
Replicate API status questions
Is Replicate down right now?
LLMPing links to the official Replicate status source and shows troubleshooting guidance. If live fetch is unavailable, use the official status page for the latest provider-owned incident data.
How do I check Replicate API status?
Start with the official status page, then run a minimal cURL request outside your app to separate provider issues from key, quota, model, network, or code issues.
Why is my Replicate API request failing?
Common causes include invalid API keys, missing model access, rate limits, quota, wrong endpoint path, CORS, provider incidents, or upstream model capacity.
What should I do if Replicate returns 429?
Check billing and quota, reduce concurrency, add exponential backoff with jitter, lower token usage, and route urgent production traffic to a fallback model or provider.
How can I monitor Replicate API outages?
The current MVP provides manual debugging tools. Join the early access list if you want server-side checks, Slack or email alerts, and historical latency reports.
Want alerts before users notice?
Join the early access list for server-side LLM API monitoring.
Phase 2 may add scheduled checks, email alerts, Slack alerts, and historical latency reports.