Why are we responsible for paying when a model burns tokens fixing 100% of its own mistakes? Shouldn't that be the model's responsibility? Some accountability would be nice.
@bcherny It’s one of those give takes. It’s clearly working for the best but clearly shows Anthropic is 100% harvesting through and keeping all our data. Just keep that data safe!
I'm not even mad anymore. I expected this. Sad because it's a great model but not usable for anything sustaining unless you're just knee deep in funding for API. At this rate, without resets, I may be able to use it once or twice more before my cancelled plan ends.