FORGE INTEGRATION FOR NVIDIA

Autonomous software delivery on NVIDIA NIM.

Forge treats NVIDIA NIM as a first-class OpenAI-compatible model provider, then adds repository intelligence, safe tools, execution and verification around inference.

Use NIM in Forge Technical integration

Integration readyBYOKHosted NIMSelf-hosted endpoint compatible

NIM

Native NIM provider

Forge includes nvidia_nim at NVIDIA's OpenAI-compatible inference base URL.

CODE

Coding-focused default

Launch default: qwen/qwen3-coder-480b-a35b-instruct, exposed through NVIDIA NIM.

GPU

Infrastructure choice

NVIDIA-hosted inference works directly; approved self-hosted NIM deployments can be connected through a custom OpenAI-compatible endpoint.

✓

Forge verification

The model proposes work; Forge runs the project, tests it and reports evidence before calling the task verified.

ARCHITECTURE

Forge → NIM → coding model

Provider: nvidia_nim
Base URL: https://integrate.api.nvidia.com/v1
Default model: qwen/qwen3-coder-480b-a35b-instruct
Endpoint: /chat/completions

NIM's OpenAI-compatible contract means Forge can reuse its hardened provider transport while preserving tenant credentials, usage measurement and verification.

COLLABORATION-READY

Designed to create measurable NIM adoption.

This is a technical integration for NVIDIA technology. No endorsement or partner status is claimed.

NIM calls can be attributed by provider
Verified coding outcomes can become benchmark evidence
Self-hosted NIM supports private enterprise deployment patterns
NGC/NPN packaging can follow formal NVIDIA approval