Issues using the Chat completion API EAST US 2 Deployments

same issue. - 504 Gateway Time-out

Same here. And the api started throwing 500 errors again as of now

Hi GS,

Thanks for reaching out. I understand you're encountering 504 Gateway Time-out and occasional 500 errors while calling the Chat Completions API in East US 2, both from the Foundry Playground and your application.

Since the issue is reproducible in both your application and the Playground, and multiple users have reported similar behavior, this strongly indicates a service-side or regional issue rather than a client-side misconfiguration.
This strongly suggests a temporary backend issue or regional degradation in Azure OpenAI (East US 2) rather than an issue with your code or configuration. Try deploying and testing the same model in a different region

That said, here are a few areas you can check to rule out common causes:

1. Regional service health

• 500 errors typically indicate server-side issues, and Microsoft recommends retrying and checking service status
• Please review Azure Service Health to confirm if there are any ongoing incidents impacting Azure OpenAI in East US 2

Since multiple users and the Playground are affected, this is the most probable root cause
2. Endpoint or deployment configuration

• Verify your request URL format:

https://<your-resource-name>.openai.azure.com/openai/deployments/<your-deployment-id>/chat/completions?api-version=2023-05-15

• Ensure:
  • You're using the deployment name (not model name)
• The deployment is in a healthy state in Azure AI Studio

3. Networking / proxy considerations

• If testing locally:
  • Try from a clean network (no proxy/firewall)
• If it works externally, allow outbound traffic to Azure OpenAI endpoints
• However, since the Playground is also failing, this is less likely in your scenario

4. Quota or throttling checks

• Azure OpenAI enforces RPM/TPM limits, and exceeding them results in 429 errors (not 500/504)

Since you're seeing 500/504, this is unlikely the root cause
5. Payload size / request complexity

• Very large prompts or tool definitions can sometimes cause latency or failures
• In similar cases, large payloads have been linked to backend errors
  Try:
• Reducing max_tokens
• Simplifying prompt or tool definitions

This strongly suggests a temporary backend issue or regional degradation in Azure OpenAI (East US 2) rather than an issue with your code or configuration. Try deploying and testing the same model in a different region

References

• Deploy & manage models:
  Azure AI Studio deployment guide
• Troubleshooting deployments:
  Deploy & monitor troubleshooting
• Quotas & limits:
  Azure OpenAI quotas
• Chat Completions API:
  Chat completions guide
• Azure Service Health:
  Check Azure service health

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.