Models we track and their prices.
Providers and gateways that host or route models.
Hardware benchmarks for running models yourself.