Ask TokensBench

Pick what you are about to do. See what similar reports actually cost, scoped and sourced honestly.

Free-text ask is coming — pick filters for now.

TaskSizeXS — One-liner or config tweak; single file; no tests needed.S — Small change in 1–2 files with a test; no design decisions.M — Reproduce, fix, and test across ≤5 files in a mid-size repo; 1–2 sessions.L — Feature or refactor touching many files/modules; design decisions involved.XL — Cross-cutting migration or greenfield build; many sessions and reviews.Harness (optional)Model family (optional)