OpenAI backend

actantial.backends.openai.OpenAIBackend

Bases: LLMBackend

Backend for OpenAI API models.

Wraps the OpenAI Responses API to provide text generation compatible with the actantial pipeline.

__init__(model_name='gpt-4o-mini', system_prompt='You are a helpful assistant.', temperature=0, api_key=None)

Initialise the OpenAI backend, validate the model, and run a validation call.

A lightweight validation call (max_output_tokens=16) is made at construction time to verify connectivity and model availability. When temperature is a float, the call also checks whether the model accepts that parameter; if not, a warning is issued and temperature is omitted from all subsequent calls.

Parameters:

Name	Type	Description	Default
model_name	str	OpenAI model identifier (e.g., "gpt-4o", "gpt-4o-mini").	'gpt-4o-mini'
system_prompt	str	System-level instruction passed to the model on every request.	'You are a helpful assistant.'
temperature	Optional[float]	Sampling temperature in [0, 2]; higher values increase randomness. Defaults to 0 for deterministic output. Pass None to omit the parameter entirely (e.g. for reasoning models that do not accept it). If a float is given but the model rejects it, a warning is issued and it is ignored.	0
api_key	Optional[str]	OpenAI API key. If None, read from the OPENAI_API_KEY environment variable.	None

generate(prompt, max_new_tokens=2048, **kwargs)

Generate text from a prompt.

Parameters:

Name	Type	Description	Default
prompt	str	The input prompt string.	required
max_new_tokens	int	Maximum number of tokens to generate.	2048
**kwargs	Any	Additional parameters passed to the OpenAI Responses API.	{}

Returns:

Type	Description
str	The generated text string, excluding the input prompt.