This says “Anthropic gives you 5× more API-priced model usage”, which is not equivalent to “Anthropic is 5x as much value”. This does not take into account model/harness efficiency. I tried looking up some benchmarks and Claude seemingly uses roughly 5x as many tokens while taking its sweet time to complete a task. No such thing as a coincidence.