Codex vs OMP harness
Reddit dev_community 上的一位用户正在质疑 OMP (Oh My Pi) 与 Codex 相比的性能。他们指出，即使在运行 10 个子代理并将所有代理的推理设置为最大值的情况下，OMP 的使用限制也出乎意料地难以达到。这让他们开始思考，通过 OMP 是否充分利用了模型的全部功能，或者与使用 Codex 相比，是否丢失了一些功能，尽管他们也承认较低的使用率并不一定意味着较差的结果。
时间与来源
时间显示为 UTC
显示时区：UTC
本地时区尚不可用，暂时显示 UTC。
发布当时偏移：UTC+02026年9月9日 21:01 UTC
收录当时偏移：UTC+02026年9月10日 00:00 UTC
- 发布
- 2026年9月9日 21:01
- 收录
- 2026年9月10日 00:00
- 来源类型
- 开发者社区
- 档位
- 社区
- 信源状态
- 正常
档位是按信源手工设定的编辑判断，不是逐条打分。
Hi everyone,
You might find this funny, but I actually have the opposite problem to y’all. My usage allowance with Astra Max feels so generous that I’m starting to wonder if the model is running at half power or something, lol.
I’ve been using OMP (Oh My Pi), and the usage limits are surprisingly hard to hit, even with 10 subagents running and reasoning set to max for all of them. That got me thinking: “Wait, is everything actually working as intended? Am I getting the full capabilities of the model through OMP, or is something getting lost compared to using Codex?”
So now I’m considering trying Codex to see whether there’s a noticeable difference.
For those who’ve used both, how does OMP compare to Codex in terms of output quality when using the same OpenAI model? Does the official Codex harness actually get better results, or is the difference mostly in workflow and tooling?
I know lower usage doesn’t necessarily mean worse results, but it’s generous enough that it made me suspicious, lol.