build1 publisher
Anthropic's $200 agent run bought 18 times the runtime of the $9 attempt
A dev.to post credits Anthropic with running the same model and the same prompt under two harnesses, 20 minutes and $9 for a broken result against six hours and $200 for a working one. The hourly spend barely moved.
Publishers:dev.to
Reality
- Evidence25
- Adoption
- Insufficient
- Hype gap+30
- Incentives45
- Confidence40