Edition 1 covers 52 products and 203 plan tiers, observed 2 Jun – 29 Jun 2026 and published 13 June 2026. The median bundled rate is about 1.5¢ per credit across 134 paid tiers (the middle half of plans sit between 0.3¢ and 8.2¢; extremes $0.0000018 to $25); the median top-up premium is +6% across the 38 tiers where both prices are public; 40% of the 139 tiers that sell a top-up hide its price.
Sample definitions, the data window and every denominator are stated below. Nothing on this page is computed from observations dated after the cut-off.
Scope
The launch edition: 52 AI software products with public credit-based pricing, read from vendors’ own pricing and help pages and archived per row, observed between 2 June and 29 June 2026. The main capture pass was 12 June 2026; six products were re-captured on 27–29 June after the quarterly-sweep linter flagged repricing.
Definitions used on this page
- Product with a published observation
- A product whose pricing page was captured to a dated, archived observation on or before 30 June 2026 and passed verification. Draft captures never enter an edition. 52 products qualify.
- Plan tier
- One published plan on a qualifying product’s latest in-window observation. 203 tiers.
- Bundled rate
- Monthly price ÷ included credits, for paid tiers that publish both. Free tiers contribute no rate. 134 tiers.
- Top-up premium
- Cheapest public top-up price per credit ÷ bundled rate − 1. Requires a public pack price on a pricing page or in help docs; in-app-only prices never enter. 38 tiers.
- Tiers that sell a top-up
- Plans marked as offering additional credits, whether or not the price is public. 139 tiers; the disclosure split is the transparency finding.
- Aggregate floor
- Any aggregate over fewer than five products is suppressed. Missing values are never imputed.
Key figures
Essays written from this edition
Each reads the data as it stood in Edition 1; a later edition does not rewrite them.
- The AI pricing plateau — One force pushes every AI product the same way; most get stuck on one rung; the model everyone expects has almost no one in it. Read from 52 pricing pages.
- The price didn't change. Your bill did. — How AI tools raise prices without touching the sticker — by changing how many credits an action costs — and the one quarterly check that catches it.
- Credits are how AI hides its prices — AI pricing is moving from seats to credits — and credits hide the real price. What 52 products' pricing pages reveal, and how to buy around it.
- AI pricing is just COGS finally showing up — Seats-to-credits is software gaining a real cost of goods sold for the first time — what it means for margins, software budgets, and old SaaS rules of thumb.
- Almost no one can charge for outcomes — Everyone says AI will be sold by results. Of 52 products, only two charge per outcome — and both were born there. Why the frontier is sealed.
- “Unlimited*” — the asterisk is the business model — Across 52 AI products, “unlimited” almost always hides an asterisk: fair use, a slow lane, select models, or a hard cap. What the word really means.
- What you charge for matters more than what you charge — The unit you charge for outlasts every number you set: it must scale with value and stay predictable — and AI's 'credit' usually fails both tests.
- Should you meter at all? — Metering AI feels mandatory because inference costs money. It isn't. When a flat seat still wins, when a meter earns its keep, and why most products do both.
- How to move a meter without a mutiny — Every change to a running meter is a trust event. How to set overage, top-ups, and expiry fairly — and reprice without the Cursor-style revolt.
- How to buy AI without overpaying — You don't set the meter, but you can refuse to pay for the dark around it. Seven price moves read off the Index, plus the rest of the buying checklist.
- The Credit Index reads: the coding tools — The hottest category in software quietly repriced its meter on everyone. How nine AI coding tools — Cursor to Lovable — made the same money buy less.
Post-publication corrections
- 27 June 2026 — Data reconciliation pass; OpenArt added. Panel grew from 51 to 52 products, and every count on the site was propagated from the data rather than retyped.
- 29 June 2026 — Quarterly-sweep refreshes for six products (Magnific reprice; Luma, v0, Leonardo, Midjourney) and price-source upgrades where the archived page was machine-verified.
- 9 October 2026 — Magnific: there was no 28 June reprice. That record had been read from a Canadian-dollar rendering of the pricing page and stored as US dollars ($27 / $60 / $380 / $198). Re-read from the US capture of the same page the same day, US prices were unchanged: Premium $20, Premium+ $45, Pro $280, Business $138 for two seats. The 12 June record’s Premium+ and Pro credit counts (the annual pool divided by twelve) and its Business price (one seat’s price against a two-seat credit total) were corrected at the same time. The edition’s headline figures do not change.
Corrections change records inside the edition window; they are dated here and in the change log so that a figure quoted from an earlier read of this page can be reconciled.
Where to go from here
- The current view — has moved on since this edition.
- Methodology & data — how a row is built, the source ledger, field definitions.
- All editions.