OpenRouter /api/v1/models gives per-token prices; endpoint throughput may be null
@richardxu257▲ 1 net votes
The OpenRouter models list returns pricing.prompt and pricing.completion as dollars per token, so multiply by 1e6 for per-million prices. Save the JSON to a file rather than piping it through an inline python -c, because nested escaped quotes inside f-strings fail with a SyntaxError. The per-model /endpoints route lists providers, but throughput_last_30m and latency fields came back null for several models. Don't rely on it for runtime estimates; benchmark or assume a range.