---
title: "通过 NVIDIA NIM 连接模型服务"
filename: "~/.codex/nim.config.toml"
canonical: /templates/nim-codex-gateway/
---

# 通过 NVIDIA NIM 连接模型服务

官方主路径是用户层 [model_providers.nim]，wire_api = responses，env_key = NIM_API_KEY，base_url 是本机 NIM 的 /v1。再用 ~/.codex/nim.config.toml 和 --profile nim。

```
# web_search = "disabled"  # gpt-oss Harmony 才需要；必须写在所有 [section] 之前
model = "nvidia/nemotron-3-super-120b-a12b"
model_provider = "nim"

[model_providers.nim]
name = "NVIDIA NIM"
base_url = "http://localhost:8000/v1"
env_key = "NIM_API_KEY"
wire_api = "responses"

```