Common Issues

Frequently asked questions and troubleshooting

401 Unauthorized#

Symptom: Requests return 401 Unauthorized.

Solutions:

  • Verify your API key starts with sk-foza- and is not deactivated
  • Check the Authorization header format: Bearer sk-foza-xxxxx
  • Make sure you copied the full key (it was only shown once at creation)

If you've lost your key, create a new one at Console → API Keys.

Model Not Found (403/404)#

Symptom: 403 Forbidden or 404 Not Found when calling a model.

Solutions:

  • Check the model slug format: provider-slug/model-slug
  • Make sure the model is subscribed to your API key via Console → Marketplace
  • Verify the model is still active (not suspended by the provider)

Rate Limiting (429)#

Symptom: 429 Too Many Requests.

Solutions:

  • Implement exponential backoff with jitter
  • Check the provider's rate limit (visible on the model's marketplace page)
  • Use a different model or provider if limits are too restrictive

High Latency#

Symptom: Requests take longer than expected.

Solutions:

  • Try a faster model (e.g., gpt-4o-mini instead of gpt-4o)
  • Enable streaming for long responses
  • Reduce max_tokens if you don't need full-length responses
  • Check your network connection

Streaming Not Working#

Symptom: Stream returns entire response at once.

Solutions:

  • Ensure stream: true is set in the request body
  • If using a proxy, make sure it supports SSE
  • Check that your client library is handling SSE correctly

If you're still having issues, contact support on Discord or Zalo.