OpenAI

The openai provider supports the OpenAI /chat endpoint.

tip

For Azure OpenAI, use azure-openai instead.

For OpenAI-compatible APIs (Groq, Ollama, Together AI, etc.), use openai-generic.

Basic Example

client<llm> MyClient {
    provider "openai"
    options {
        api_key env.MY_OPENAI_KEY
        model "gpt-4o"
        temperature 0.1
    }
}

BAML-Specific Options

Option	Type	Default	Description
api_key	string	env.OPENAI_API_KEY	API key for Authorization header
base_url	string	https://api.openai.com/v1	Base URL for API
headers	object	-	Additional request headers

Custom Headers

client<llm> MyClient {
    provider openai
    options {
        api_key env.MY_OPENAI_KEY
        model "gpt-4o"
        headers {
            "X-My-Header" "my-value"
        }
    }
}

Available Models

Model	Use Case	Context	Features
gpt-4o	General purpose, multimodal	200K	Updated knowledge, June 2024
gpt-4o-mini	Cost-effective tasks	128K	Faster, lower cost
o1	Complex reasoning	200K	Extended thinking
o1-mini	Lightweight reasoning	128K	Faster reasoning

See OpenAI Models for the complete list.

Reasoning Models

warning

For reasoning models (o1, o1-mini), use max_completion_tokens instead of max_tokens:

client<llm> OpenAIo1 {
    provider openai
    options {
        model "o1-mini"
        max_tokens null
    }
}

Changing Regions

Access OpenAI's API in different regions by setting base_url:

client<llm> EUClient {
    provider openai
    options {
        model "gpt-4o"
        base_url "https://eu.api.openai.com/v1"
    }
}

Provider Parameters

These options are passed directly to OpenAI:

Option	Type	Description
model	string	Model identifier
temperature	float	Sampling temperature (0-2)
max_tokens	int	Maximum response tokens
top_p	float	Nucleus sampling parameter
frequency_penalty	float	Frequency penalty (-2 to 2)
presence_penalty	float	Presence penalty (-2 to 2)

See the OpenAI API documentation for all options.

See Also

• Azure OpenAI
• OpenAI Generic
• Client Overview