They can. Results change as you change your models, and results aren't always strictly better or worse, which is why testing gold-standard results with any prom

They can. Results change as you change your models, and results aren't always strictly better or worse, which is why testing gold-standard results with any prompt and model changes is so important for applications utilizing LLMs.