How to Get the Lowest-Cost LLM Inference on OpenRouter

Append :floor for the cheapest provider, cap spend with max_price, and start free with 20+ zero-cost models. Plus the billing gotchas to avoid.