Post

I find using different core models more useful than just picking one "best" one. It's like getting different perspectives. They all find things the others miss and approach a problem differently. Models from the same family or with different prompts are too similar.