NVIDIA Nemotron API returns HTTP 403 although model listing succeeds

AI & Data Science NVIDIA NIM Models
1

Hello NVIDIA Developer Support,

I am integrating the NVIDIA Nemotron model into an AI project running on AWS EC2.

My API key successfully authenticates against the model catalogue, but inference requests are rejected.

  • Model: nvidia/nemotron-3.5-lightning-30b-a3b

  • Base URL: https://integrate.api.nvidia.com/v1

  • GET /v1/models: HTTP 200

  • POST /v1/chat/completions: HTTP 403

  • Error response: {"status":403,"title":"Forbidden","detail":"Authorization failed"}

I have verified that the model appears in the model catalogue and that the request is being sent to the documented endpoint.

Could you please check whether my NVIDIA Developer account or organization has the necessary hosted inference / Public API Endpoints permissions, and advise how to enable access?

I can provide the account details and request timestamps privately if needed.

Thank you.