Skip to Content

LLM Configuration

Connect Mixtral

Use a Mixtral model on any host with an OpenAI-compatible API for Bryn, copilot and natural-language questions.

Before you start

  • Your role has LLMs > Edit. Admins have it.
  • The base URL of the OpenAI-compatible API, for example https://<your-provider>.ai/v1, and its API key.
  • Databrain can reach the URL. On Databrain cloud, the endpoint must be on the public internet.

Steps

  1. Go to Settings > LLMs and select Add LLM.
  2. Select Mixtral.
  3. In Name, enter a name, for example Mixtral for Bryn.
  4. Enter the Base URL and the API Key. Databrain checks them once both are complete.
  5. If the server does not list its models, enter the Model names: the full model name your provider expects.
  6. In Output limit parameter, pick the field your endpoint reads for the output limit:
    • Both: safe for Ollama.
    • max_completion_tokens: for an OpenAI pass-through.
    • max_tokens: for endpoints that read only this field.
  7. Select Authenticate.
Connect Mixtral dialog with the Name, Base URL and API Key fields, the Output limit parameter set to Both, and the Authenticate button
Connect Mixtral with Name, Base URL, API Key and Output limit parameter.

The LLM shows as a card in Settings > LLMs with its models.

If something goes wrong

  • The check fails. Make sure the base URL ends with the API version path (often /v1) and that the key is valid.
  • Answers stop early or the provider refuses the request. Change Output limit parameter to the field your endpoint reads.
  • No models show. Enter the Model names by hand.

Next, set the default model.