Mixing models isn't about hype. You use them to hedge against specific failure...

https://fast-wiki.win/index.php/The_Synthesis_Trap:_How_We_Are_Laundering_Dissent_Out_of_AI_Outputs

Mixing models isn't about hype. You use them to hedge against specific failure modes and leverage niche strengths. But watch out. Merging outputs can mask important dissent or leak data across providers. Build for blind spots, not perfection.