Prices are per million tokens; the cheapest route is shown in the model view. Uptime, throughput, and latency come from each route's labeled stats source. Unknown figures show a dash, never $0.