403 Authorization failed on integrate.api.nvidia.com — Public API Endpoints unavailable

Hi NVIDIA team,

I’m unable to make inference calls through the NVIDIA hosted API.

My situation:

What works:

  • GET /v1/models → HTTP 200
  • NVIDIA Playground → model responds successfully
  • API key is freshly generated and loads correctly

What fails:

  • POST /v1/chat/completions → HTTP 403
  • Response:
    {“status”:403,“title”:“Forbidden”,“detail”:“Authorization failed”}

I have tested the request using both Python requests and the OpenAI-compatible Python client.

I also checked my Organization Details, but I cannot find any setting or permission called “Public API Endpoints”.

Could you please check whether Public API Endpoints/inference access is enabled for my Personal organization, or advise what needs to be enabled to use the hosted inference API?

Thank you.