Cheaper tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → do not necessarily make an AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → task cheaper. In Artificial Analysis' September evaluation of Claude Opus 5.5, Anthropic's lower prices accompanied substantially more generated output, leaving average cost per task roughly level with Opus 5 at maximum effort.
The report also evaluated multiple effort settings, which trade additional computation against measured capability. That makes effort level part of the comparison rather than a minor configuration detail. The release's value cannot be read from its per-token discount alone: total usage, cache pricing and the quality needed for the task determine whether the change helps a particular workflow.
Terms in this piece · Glossary
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
