NVIDIA NIM × FORGE

OpenAI-compatible NIM inference, wrapped in Forge verification.

Use NVIDIA NIM without learning another Forge mode. Select NIM—or let routing choose it later—and keep the same plan, build, test, repair and verify workflow.

Hosted NIM

provider = "nvidia_nim"
base_url = "https://integrate.api.nvidia.com/v1"
model = "qwen/qwen3-coder-480b-a35b-instruct"

Forge sends OpenAI-compatible chat-completion requests with the customer's NIM API key.

Self-hosted NIM

For an enterprise NIM running inside approved private infrastructure, configure its OpenAI-compatible base URL through Forge's custom provider path and keep the same agent workflow.

/v1/chat/completions compatible
Tool calling supported where the chosen NIM supports it
Workspace policies remain enforced by Forge
BYOK / private endpoint ownership stays with the customer