AI Providers

Supported providers#

The current release ships one built-in provider: NVIDIA, using the NVIDIA NIM API — an OpenAI-compatible endpoint at https://integrate.api.nvidia.com/v1. If the configured provider is missing or invalid in your config, Dash falls back safely to its defaults.

Connect a provider#

Inside the TUI:

/providers
# switch to a different model list:
/models

• /providers (alias /provider) opens the API key popup where you paste your nvapi_… key.
• The key is stored in ~/.dash/config.json under providers.nvidia.apiKey, encrypted with AES-256-GCM.
• Every API request is sent as Authorization: Bearer <key> to the NVIDIA NIM endpoint.

Choosing a model#

/models opens the model popup. Dash fetches the live model list from the provider (GET /models) when a key is configured. If the request fails or no key is set, Dash falls back to built-in defaults:

[
  "nvidia/nemotron-3-ultra-550b-a55b",
  "nvidia/nemotron-3-super-120b-a12b"
]

The selected model id is written to providers.nvidia.model in your config.

Working without a valid key#

• With no key, generating answers fails with a reminder: Set API key first with /providers.
• With an invalid key, provider HTTP calls fail — see Troubleshooting.