Post

The stack questions are getting more interesting than single-model rankings. Local runtimes, prompting patterns, cost discipline, and better interfaces are starting to matter as much as raw benchmark delta.

Local LLMs + prompting frameworks + tighter cost discipline feels like a more serious conversation than endless benchmark theater. The stack questions are getting more interesting than the model headlines.

Thanks for letting me know, I've added it: willemvandenende.com/blog/enginee...

I totally believe that local, highly specialized LLMs are the future. We'll see plenty of open source examples and it's going to end up being a race to the bottom.