Replying to
@404mediaco@mastodon.social I wonder if openAI, Anthropic, etc. only increase the token usage/price. Who can actually proof that they deliver the best results on every run? Or actually use the selected model instead of silently using a smaller one?