#How can I use the NVIDIA API key?

1 messages · Page 1 of 1 (latest)

warped islandBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

heady fulcrumBOT
#

You can do this as a custom OpenAI‑compatible provider (NVIDIA’s Kimi endpoint is OpenAI-ish).

From NVIDIA’s own example, the endpoint is:

  • Base URL: https://integrate.api.nvidia.com/v1
  • Model id: moonshotai/kimi-k2.5
  • Auth header: Authorization: Bearer $NVIDIA_API_KEY

Option A (recommended): rerun the wizard and pick “Custom provider”

On the VPS run:

openclaw onboard

In Model/Auth, choose Custom provider (compatibility: OpenAI), then enter:

#
  • Base URL: https://integrate.api.nvidia.com/v1
  • Model ID: moonshotai/kimi-k2.5
  • API key: your NVIDIA_API_KEY (don’t paste it here in Discord)

Option B: edit config directly (~/.openclaw/openclaw.json)

  1. Put the key where the daemon can read it (best place):

~/.openclaw/.env

NVIDIA_API_KEY=nvapi-...
  1. Add a provider + set it as default model:
{
# 
  models: {
    mode: "merge",
    providers: {
      nvidia: {
        baseUrl: "https://integrate.api.nvidia.com/v1",
        apiKey: "${NVIDIA_API_KEY}",
        api: "openai-completions",
        models: [{ id: "moonshotai/kimi-k2.5", name: "Kimi K2.5 (via NVIDIA)" }],
      },
    },
  },
  agents: {
    defaults: {
      model: { primary: "nvidia/moonshotai/kimi-k2.5" },
    },