期待能给出满意的答复啊 没有用任何sub2api 类似的，gmail邮箱。老号，原来免费，这一月订阅菲区
就是sol xhigh 一直用了估计2天（就是到中国时间0814 18:00 估计不到2天）， 另一个会话sol xhigh用了10h，现在周限0%了。
体感上降额了
但是用论坛里油猴脚本测试正常啊
————————回复——————-
Thank you for confirming that Fast mode was not enabled.
I can see why this usage feels surprising, especially when you were expecting your Pro weekly allowance to last longer. I can clarify what likely contributed to the usage and what your current Codex status shows.
Since Fast mode was not enabled, that likely was not the cause here. The main driver is that GPT-5.6 Sol with xhigh reasoning can use your Codex weekly allowance very quickly, even without Fast mode.
Codex usage is not measured only by how many hours a session runs. It depends on how much work the model performs, including repo size, context length, tool calls, file reads, retries, long-running tasks, and how much conversation history the session is carrying.
In your case, you had one Sol xhigh session running continuously and another Sol xhigh session running for about 10 hours. Those sessions both count toward the same Codex allowance on your Pro account, so they can add up quickly. A 10-hour Sol xhigh run on a large or complex codebase can consume a large part of the weekly limit by itself, and a continuously running session can continue using more allowance as it works through tasks.
To reduce usage going forward, I recommend:
- Stop or pause any Codex sessions that are not actively needed.
- Start a fresh session for a new task instead of keeping a very long session running.
- Use GPT-5.6 Terra or Luna for simpler edits, debugging, or routine coding work.
- Use Sol xhigh only for the most complex tasks that need deeper reasoning.
- Keep the working context smaller when possible, especially with large repositories.
Based on what you described, one day of heavy Sol xhigh usage across two sessions can plausibly consume most of the weekly Codex allowance. This does not necessarily indicate an account or billing error.