model routing

Flash-Tier LLMs Are the Real Inference Efficiency Story

Here's a claim that will annoy a lot of people who spend their days arguing about model leaderboards: for most production LLM workloads, the "smartest" model...