Skip to content
1 minute readNewAgent-ready

See which pricing function suits which kind of request

A new table shows quality, speed and cost for each kind of request (short questions, long text, code, structured data, other languages) and each pricing function.

Requests differ, and so do pricing functions. This table tells you which function answers which kind of request best, fastest and for the least.

  • Each request is put in one class by fixed rules on its free text, with no AI model. The text is never stored, only the class
  • Benchmark rows come from a test bench of made-up requests with a known best price, and are marked as such
  • Rows for your own workspace appear once you have enough routed calls of a kind, and only your workspace's calls are read
  • Read it at GET /api/routing/prompt-class-benchmarks, or from your agent with the get_prompt_class_benchmarks MCP tool