Your agent solved the task at xhigh. Would low have solved it too? You can’t tell without rerunning it, and you still pay for every token. We tested models x reasoning effort level x benchmarks. On tasks low effort can already solve, top effort costs up to 42× more #LLM #AI 🧵👇 ...
