Expand description
NVIDIA provider.
Targets NVIDIA’s hosted inference API at https://integrate.api.nvidia.com,
which exposes the OpenAI /v1/chat/completions shape with Bearer-token auth
(nvapi-... keys). Apart from the endpoint, env var, and default model list
it is identical to the cloud openai provider, so it reuses the shared
openai_compat request/response/streaming/tool helpers.