Pricing data

This page is about the MODEL RATE CARD — the provider prices we use to price your usage. (Our own subscription prices live on the pricing page and vary by region.)

Money math is only as good as the rate card behind it. Ours is versioned, effective-dated, and verified by a human — deliberately, as a trust feature: no scraped or "live" price can silently change your audit's dollars — and the report tells you exactly which version priced your audit.

Four rates per model

Every model carries four USD-per-million-token rates: input, output, cache write, cache read. Modern waste analysis is impossible with a two-rate card — cache economics (the largest waste class in agent traffic) live entirely in the write premium and the read discount.

Effective-dated, timestamp-matched

Rates carry effective_from dates, and each call is priced at the rate in effect at its timestamp. When a provider changes prices mid-window — like Sonnet 5's introductory pricing ending 2026-08-31 — a log spanning the boundary prices each side correctly instead of smearing one rate across both.

Model keys match exactly, or as dated snapshots (model-key-2...), longest key first — gpt-5.4-nano can never silently take gpt-5.4's card.

Strictly verified, on the record

The table carries a last_verified date — the last time our strict verification gate ran green. On every release, an automated verifier compares every current rate row against independent published price data; one mismatched or uncorroborated row and the release does not ship. There is no override and no human bypass. Our CI warns loudly when the date exceeds 14 days; the founder's daily ops digest carries the age.

A daily sync additionally tracks routine rate updates from the same independent data, and large swings are held and surfaced rather than silently applied. Both paths log what they compared and what changed — the record is the machine's, checkable by anyone.

Why we refuse live pricing

Scraped or API-fetched rates change without audit trail and break reproducibility: the same log re-audited next month must price identically unless a human recorded a rate change. A versioned file with dated rows is slower to update and much harder to be wrong with.

When a model is not in the table

Calls on models without a verified rate card are excluded from totals and listed in the report as unpriced — count and model ids — rather than priced by guesswork. Spend reconciliation (our ±0.5% property test) runs only over priced rows and says so.