Skip to content

MedGemma as API service on Google cloud #20

Description

@erangross27

Hi everyone,
I'm looking to use a model through API requests—like we commonly do these days—when local deployment isn’t feasible. I'm currently developing a new healthcare platform on my dev machine, and the underlying model is Gemma4B. However, with only an 8GB Nvidia GPU, it takes around 12 minutes to get a single response, which is far too slow.
Instead of running models locally, I want to make inference calls using an API key—like we used to do in the good old days. I tried Hugging Face, but wasn’t able to purchase credits for their inference service.
I'm not interested in setting up a Google Cloud endpoint or doing fine-tuning. I simply want to tap into a model’s existing knowledge to analyze symptoms from text or images as part of a comprehensive healthcare platform.
Thanks,
Eran

Metadata

Metadata

Assignees

No one assigned

    Labels

    type: supportUser is asking for help / asking an implementation question.

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions