Specify model settings when using google generative AI

[tiny ocean]

Is it possible to directly call an Gemini model with custom settings in an automation? For example if I want to specify what model to use, temperature, etc?
I notice that the OpenAI automation lets you specify what config entry to use, but there's nothing like that for gemini. Is there no way to edit those settings?
If I have multiple config entries with different models, which model does it use?

[violet remnant]

look at using conversation.process which is the generic home assistant action. instead of the integration specific action.

you may not get some functionality provided by the specific integration action but if you only need text then you likely wont have an issue with the generic action.

[tiny ocean]

Yep, that does indeed work for specifying the model at least, though that does make me wonder what model is even being used here with Google generative AI. Is it the first one? Or random? I'm really not sure

I do wish there was an action for power users to directly call the models with custom settings and stuff

[violet remnant]

as of 2025.7 google generative ai uses Gemini 2.5 Flash by default. but you can change it in the integration settings

[tiny ocean]

Whereabouts would you actually change that? I can set it for an individual conversation agents

[violet remnant]

yeah thats what i meant sorry, you set it for the agents then you can have multiple agents of different models if you want. then you select desired agent when using conversation.process

[tiny ocean]

Oh sorry yeah that makes sense
So which model does it use when you use google_generative_ai_conversation.generate_content? I have multiple conversation agents, some with flash, others with pro

For example I want to use the pro model here in my script

[violet remnant]

I am not sure on that, its not an option i am familiar with. hopefully someone with more knowledge of that part can jump in at this stage.