100+ AI models from 20 providers. Prices per 1M tokens.
💡 Cached input = reused context tokens (typically 50-75% of input price)