Claude 额度没涨，只是模型单价更便宜
分析称 Codex 原始额度仍高于 Claude，变化在于 Astra 每百万 token 10/50 美元比 Opus 5.5 的 4/20 更贵，同样上限下烧得更快。
🚨 Claude's usage limits didn't actually go up it just feels that way **The overall numbers:** Codex still hands out more raw quota than Claude. That hasn't changed. **What did change:** the pricing gap between models. - GPT-6 Astra: bigger, pricier — $10/$50 per million tokens. Every task burns through more of your allowance. - Opus 5.5: leaner and cheaper — $4/$20 per million tokens. Same usage cap, but it stretches much further. **The takeaway:** your limit didn't move. How fast you burn through it did. That's the real trick — Astra-level intelligence (or better) packed into a smaller, cheaper model.
这条是引用着另一条写的：x.com/codexresets1/status/2104147350444802121