LLMs: Intelligence vs. cost
Summary
OpenTeams critiques the Intelligence vs. Cost benchmarking approach used by ArtificialAnalysis, arguing that log-scale cost charts can obscure meaningful price differences and that open-weight models or local deployments are often misrepresented. The article presents alternative plots on linear cost scales, discusses local vs datacenter costs, and provides guidance on choosing models based on practical use-cases and total cost of ownership.