Using NVIDIA NIM with OpenClaw

Prerequisites

• ✓An NVIDIA account with access to API keys at https://build.nvidia.com/settings/api-keys.
• ✓An existing OpenClaw installation with the `openclaw` CLI available in your shell.
• ✓Network access from your OpenClaw environment to `https://integrate.api.nvidia.com/v1`.

Steps

1. 1

   Create an NVIDIA API key

   Start by creating an API key in your NVIDIA account so OpenClaw can authenticate against the NIM endpoint. You do this in the NVIDIA web UI, and you will use the resulting key as an environment variable in later steps.

   textCopy

   Create an API key at https://build.nvidia.com/settings/api-keys.

2. 2

   Export your NVIDIA_API_KEY and run OpenClaw onboarding

   Set the `NVIDIA_API_KEY` environment variable in your shell so OpenClaw can auto-enable the NVIDIA provider. Then run the onboarding command; using `--auth-choice skip` keeps onboarding focused on local config without extra auth prompts.

   bashCopy

   export NVIDIA_API_KEY="nvapi-..."
   openclaw onboard --auth-choice skip

3. 3

   Set an NVIDIA model as the active model

   Tell OpenClaw which NVIDIA model to use by default. This command registers `nvidia/nvidia/nemotron-3-super-120b-a12b` as the current model for your workflows, so subsequent agent runs will target this model unless you override it.

   bashCopy

   openclaw models set nvidia/nvidia/nemotron-3-super-120b-a12b

4. 4

   Configure NVIDIA as a provider in your OpenClaw config

   If you maintain an explicit config file, add the NVIDIA provider block so your setup is self-documenting and easy to version-control. This config points OpenClaw at NVIDIA’s OpenAI-compatible `/v1` endpoint and sets the default agent model to the Nemotron 3 Super 120B reference.

   jsonCopy

   {
     env: { NVIDIA_API_KEY: "nvapi-..." },
     models: {
       providers: {
         nvidia: {
           baseUrl: "https://integrate.api.nvidia.com/v1",
           api: "openai-completions",
         },
       },
     },
     agents: {
       defaults: {
         model: { primary: "nvidia/nvidia/nemotron-3-super-120b-a12b" },
       },
     },
   }

5. 5

   Choose a model from the built-in NVIDIA catalog

   OpenClaw ships with a static catalog of NVIDIA-backed models you can target by reference. Pick the model ref that matches your use case and swap it into your `openclaw models set` command or config file.

   textCopy

   | Model ref | Name | Context | Max output |
   | --- | --- | --- | --- |
   | `nvidia/nvidia/nemotron-3-super-120b-a12b` | NVIDIA Nemotron 3 Super 120B | 262,144 | 8,192 |
   | `nvidia/moonshotai/kimi-k2.5` | Kimi K2.5 | 262,144 | 8,192 |
   | `nvidia/minimaxai/minimax-m2.5` | Minimax M2.5 | 196,608 | 8,192 |
   | `nvidia/z-ai/glm5` | GLM 5 | 202,752 | 8,192 |

Configuration

Troubleshooting

export NVIDIA_API_KEY="nvapi-..."

export NVIDIA_API_KEY="nvapi-..."

Frequently asked questions