Claude Code's hidden effort_cost_index table reveals max tier costs vary wildly by model
Wsz2020 · reddit · 2026-09-18
A hidden effortcostindex table in the Claude Code binary shows per-model token cost per effort level (high=1). Sonnet 5's max tier costs +132% over xhigh, while Opus 5 flattens to just +6% — "max" means very different things per model. The UI string confirms costs are per attempt, not per success, with no retry pricing. The author's own 8-agent measurement (2 tasks x 4 levels on Opus 5) confirmed the direction and nailed the key prediction: xhigh and max output tokens were nearly identical (397,827 vs 397,761). No documentation exists on how the catalog numbers were derived — treat as vendor estimates.
More from coding & agent
- Butterfly traces multi-agent failures with graphs, 62.5% accuracy vs 53.8% for log prompting — JeremyCMorgan · 2026-11-04
- Free course covers the 4 hard problems of production generative media agents — Al_Grigor · 2026-11-04
- Quota reset night: Codex back to 100% while Claude shows just 10% left — rudrank · 2026-11-04
- Self-improving software is inevitable: closing the agent-observability loop — bibryam · 2026-11-04
- Building a 7-GPU 132GB VRAM local inference rig: consolidate on EPYC or split across two systems? — infectiousstupidity · 2026-11-04
- Founder of 50-person, 12-country team says he talks to AI agents more than humans — Scobleizer · 2026-11-04