API-equivalent cost
API-equivalent cost is what a session's token usage would cost at a provider's list API prices, a yardstick for comparing models and workloads when the work actually ran on a subscription that does not bill per token.
It is computed line by line from the usage an agent reports: fresh input, cache writes, cache reads and output, each multiplied by the model's published price. Output includes thinking tokens, which are billed as output. On a flat-rate plan none of those dollars are charged; the figure says what the same traffic would have cost on an API key.
Claude Code displays one. As of September 2026, the session block of its /usage screen shows a dollar amount computed locally from token counts at list price, and Anthropic's documentation notes that for Pro and Max subscribers this figure is not relevant for billing purposes.
Its value is comparison. Raw token counts cannot rank two models, because their prices per token differ and so can their tokenizers. Dollars at list price put a Haiku run and an Opus run on one scale, with caching and output already accounted for, which makes the figure the natural basis for cost per task.
It is neither a bill nor a quota. Subscription usage limits are not expressed in these dollars, contracted API rates can differ from list prices, and the figure is only as current as the price list behind it, so a published API-equivalent cost should carry the date of the prices it used.