What model does Gemini Advanced use?

[toddw777]

Hello,

We're using the Gemini API to process PDF files. We're generally using Gemini Advanced (in a browser) to refine prompts; however, we're getting different results when using the Gemini API. We're using the gemini-1.5-flash-001 model, which seems to work best, but getting different (and even inferior) results from what we get when using Gemini Advanced in the browser. Is it possible to use the same model in the Gemini API that is being used by Gemini Advanced so that we get consistent results?

Thanks!

[AndrewB]

Gemini Advanced now uses various models and so you won't be able to replicate exactly how you are describing. Some things to try are: 

• Experiment with Temperature and Top-P: These parameters control the randomness and creativity of the model's output.

•  Consider Fine-Tuning: If you have a substantial amount of data and specific requirements, consider fine-tuning a Gemini model on your own data. This can lead to even greater consistency and performance tailored to your use case.

• Try a larger model like 1.5 Pro, but consider the tradeoffs of speed and cost. 

[Huule]

What do you mean various models of gemini advanced? Isnt it gemini 1.5 pro?

[AndrewB]

While Gemini Pro 1.5 is the foundation model, Gemini Advanced as a product is maturing and additional features like Gems, image creation, etc. and so as the user, you are more abstracted from what is happening under the hood (things like pre-processing).

So what I mean is that I wouldn't consider it a 1:1.

Another idea is to try is additional extraction.