OpenAI simply doesn’t get it

I’m often asked whether I prefer Claude or Codex, and honestly, I’ve never really shown any preference towards either. Both seem to have reasonably good levels of intelligence, so I’ll go with whatever is more affordable. For me, that has historically been Codex with their Pro plan. I work a lot with Codex, whether I want to or not. I prefer using it as part of my IDE experience, within a tool like PhpStorm using the integrations JetBrains has built-in to allow the seamless bridge between the native, deterministic intelligence that it provides with the agent now piloting it. Sometimes I’ll use the CLI, but rarely if ever will I reach for the desktop app. Why? It’s clunky. jarring, and doesn’t really know what it wants to be. There’s nonsense around integrations, pets, projects. There’s Codex, but there’s also ChatGPT, and there’s Work and Chat, and one uses your limits whereas the other doesn’t. It’s a hodgepodge of the worst possible design decisions and using it is a bother.…

Commenting on Bubbles has been removed. Why? If you want to talk about this post, write to the author, use the comments on their blog if it has them, or reply with a "RE:" post on your own blog. Suggestions? feedback@bubbles.town