load

Contents

load#

rapidtools.models.load(provider: str, **kwargs: Any) → BaseInferenceModel[source]#

Instantiate a model wrapper from its registry key.

Keyword arguments are passed to the wrapper constructor, e.g. api_key and model_id for hosted providers, model_id, device and load_in_4bit for local ones.

Parameters:

  • provider – Registry key (see list_providers()).

  • **kwargs – Constructor arguments of the wrapper.

Returns:

BaseInferenceModel – The ready-to-use model.

Example

>>> from rapidtools.models import load
>>> gemini = load('gemini', api_key='AIza...')
>>> qwen = load('qwen_vl', model_id='Qwen/Qwen3.5-9B', load_in_4bit=True)
>>> qwen.run_inference('roof.jpg', 'Describe the roof.').text
Contents