Using multiple LLMs is tempting for solving edge cases, but it adds serious...

https://technivorz.com/the-hidden-tax-of-multi-model-architectures-why-more-models-often-means-less-intelligence/

Using multiple LLMs is tempting for solving edge cases, but it adds serious operational debt. While mixing models lets you leverage specific strengths or cross-check output for better accuracy, it is not a silver bullet