Turbo model vs davinci model use cases

Is there any documentation regarding the use case of the two models?

So far I read that turbo model cannot be fine-tuned.

How about specific biases? Is there any difference?