Hacker News new | past | comments | ask | show | jobs | submit
They are not artificially cheap, they are still cheap even when hosted by independent inference providers. Are all providers subsidizing their open-weight models?