Credit pricing is starting to crack

Metered pricing has been the default in this category for two years. In the last quarter it has started to move.

Three of the eleven vendors in our index changed how they meter agent work. Two introduced a configurable spend ceiling, which our reviewers had been asking for since we started tracking real spend. One reduced the cost of retries after a failed generation, which is the single change most likely to affect what buyers actually pay.

Why the retry change matters most

Our cost tracking found that the gap between a good month and a bad month is almost entirely about failed attempts. A feature that costs 6 EUR when the agent lands it first time can cost 30 EUR when it does not, and the buyer has no way to predict which they will get.

Charging less for a retry after a failure shifts some of that risk back to the vendor, which is where it belongs, since the vendor controls whether the agent succeeds.

What has not changed

Nobody publishes cost per completed feature on a reference brief. Until they do, comparing metered plans is guesswork, and our value axis is doing that work on the buyer's behalf.

The two flat-priced platforms in our index have not moved, and our reviewers who do fixed-price client work continue to choose them for that reason alone.

What to watch

If the spend ceiling becomes standard, the practical difference between metered and flat pricing narrows to a forecasting problem rather than a risk problem. That would be good for buyers and would probably reduce how much our value axis moves between quarters.

We re-verify pricing on every builder quarterly, and the last verified date on each review page tells you how fresh the number you are reading is.