As the part owner of a company using local LLMs we're getting real value out of 27b sized models. I'm saving $1000/mo vs throwing a frontier model at a rather s

As the part owner of a company using local LLMs we're getting real value out of 27b sized models. I'm saving $1000/mo vs throwing a frontier model at a rather straightforward (for AI) automation need. The trick is acknowledging they're B or C tier intelligence and speed and using them accordingly.