Codex quota cut debate: Astra usage effectively limited
The author argues this is not a simple 50% cut but the end of free Astra tokens, and that OpenAI is doing much the same as Anthropic's Fable limits, just framed differently.
I think people who are looking at this as a straight up 50% slash in codex usage are either confused or purposely being obtuse. For the models that are meant to be your daily drivers on a subscription plan, they've already reduced the price by half, so recalculating for that makes sense. I think the biggest implication here is no more free lunch for Astra tokens. Btw, this is EXACTLY what Anthropic does, but clearly OpenAI doesn't want to frame it like that. They are more or less imposing the same 50% usage limit for Astra tokens on your sub as Anthropic does for Fable tokens, only the semantics are different. They are saying, you can use all Astra tokens on your sub, we aren't limiting you that way, but you won't get as much usage. You shouldn't be using Astra for everything anyway. All that said, this is still a terrible look for OpenAI and the explanation here was even worse by Tibo. Maybe if these services would just say what they mean and stop treating literally everyone reading like a negative IQ reset beggar, this wouldn't have been taken so poorly.
This post quotes another post: “OpenAI reopens $200 Pro subscription with a new usage calculation”