OpenAI

The openai sub-block configures the OpenAI cloud provider. It also works with any OpenAI-compatible endpoint, such as Azure OpenAI or a local proxy, by setting config.base_url.

Fields

The config block is optional. When omitted, the OpenAI SDK reads credentials from the OPENAI_API_KEY environment variable using its standard resolution logic.

params

The params block sets inference parameter defaults for all models under this provider. All fields are optional and support runtime().

Example

providers {
  openai {
    models = [
      "gpt-4o-mini",
      {
        name = "gpt-4o"
        params {
          max_tokens  = 8192
          temperature = 0.3
        }
      }
    ]

    config {
      api_key  = secret(runtime("OPENAI_API_KEY"))
      base_url = runtime("OPENAI_BASE_URL", "https://api.openai.com/v1")
    }

    params {
      max_tokens        = 2048
      temperature       = 0.5
      frequency_penalty = 0.1
    }
  }
}

In this example, requests to gpt-4o-mini use the provider-level defaults. Requests to gpt-4o use its own model-level params for max_tokens and temperature, while still inheriting frequency_penalty from the provider-level defaults.