The cheapest plan for both Claude and Codex is the sweet spot IMO.
replyIt also forces you to keep your workflow mostly harness-independent because Claude supports next to no standards and Codex does some.
That works until openai does the same thing. Pretty clear as an industry they want to establish a new price floor for non-trivial coding use.
replyThank god for the Chinese labs. Keeping us (relatively) honest.
replyYep, and the price point theyre looking at is 95% of an engineer.
replyOnce they get people hooked, deskilled, and paying, the money ratchet only tightens.
And the companies KNOW that theyre replacing engineers, or trying to. So each engineer replaced is X salary a year they now have available, so make it back in SaaS LLM tokens.
They confirmed that it does not affect existing users
replyThat's what Claude is testing I guess (people often don't do what they say they do when it comes to pricing)
replyPresumably for new subs.
reply