Original Summary

came up with some shorthand for codex models and level:<p>low--- gpt-5.6-luna low<p>low-- gpt-5.6-luna medium<p>low- gpt-5.6-luna high<p>low gpt-5.6-terra medium<p>low+ gpt-5.6-terra high<p>low++ gpt-6-luna medium<p>med gpt-6-luna high<p>med+ gpt-5.6-sol high<p>med++ gpt-5.6-sol xhigh<p>high gpt-6-sol high<p>high+ gpt-6-sol xhigh<p>high++ gpt-6.1-sol high<p>xhigh gpt-6.1-sol xhigh<p>xhigh+ gpt-6.1-sol max<p>xhigh++ gpt-6-astra max<p>xhigh+++ gpt-6-astra ultra<p>and this script:<p><a href="https:&#x2F;&#x2F;gist.github.com&#x2F;andrewarrow&#x2F;36a7a0bb062d49709f8a656067ab58cd" rel="nofollow">https:&#x2F;&#x2F;gist.github.com&#x2F;andrewarrow&#x2F;36a7a0bb062d49709f8a6560...</a><p>Examples:<p><pre><code> dex1 med dex1 high+ &quot;Fix the failing tests&quot; dex2 xhigh+++ dex1 xhigh--- Same rung as high</code></pre>

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/10/1 20:54:19