Most teams use multi-model setups to hedge against individual model failures...

https://lima-wiki.win/index.php/What_Does_It_Mean_That_Models_Fail_Differently%3F

Most teams use multi-model setups to hedge against individual model failures and lean into specific strengths. While comparing outputs is smart, synthesis often hides vital dissent or creates cross-provider privacy leaks