finishReason

Returns why the model stopped generating.

If there are multiple candidates, this returns the reason from only the first one. It is null while a streamed response is still in progress, and set on the chunk that ends it.