Why is there such a big difference in answers between AI studio and Vertex AI

I’ve noticed significantly higher quality answers from AI studio as opposed to Vertex AI when trying to summarize utility bills and it’s understanding of the relevant PDFs

Would anybody know why? The prompts and models are the same, but the outputs vary very wildly.

Specifically, for a phone bill, Google AI studio has been able to understand why my bill decreased when I stopped two additional testing lines of service, but Vertex AI has not been able to.

Welcome to the dev forum @obsessiveorange

Were the requests completely identical?