Anthropic Obfuscates Billing Calculus
Anthropic’s pricing tier triggers as it relates to one’s consumption of API and subscription services intentionally obfuscate consumption calculus. It is difficult to calculate spend when in the midst of a build, which obviously triggers usage levels that exceed the upper bound of normal, steady-state usage. It is shit policy from a cult company. “Everyone at Anthropic’s number 1 focus is AI safety” says Dario, says Boris Cherny, say others. First, that is bullshit. Second, how about billing transparency as an operating principle? I never have usage issues with OpenAI, regardless of what tier I may be on in a given month.
I would pop over to OAI’s Codex 5.5 now, but I hate to switch models during a build. We finished the earnings call sentiment transcription engine and dashboard build, now it is just a question of moving the product off of our local servers into Google Cloud, and then building an API to extend it into Kilby so that it may be both a standalone T2D product and a Kilby “feed”.