OpenAI Codex + GPT-5.5 Is Really Good Now
The latest Codex and GPT-5.5 are worlds better than the last time I tried them - and built-in browser verification is the most important skill in my setup. It may be a mistake to standardize on one platform right now.
The latest OpenAI Codex and GPT-5.5 together are really good for coding workflows.
The Codex app even builds in probably the most important skill in my setup which verifies the work done in browser after it's done.
It's worlds better than the last time I tried it and may become my daily driver if the combination of Claude Code + Opus 4.7 keeps going down + burning through usage limits at the rate it has been vs. essentially the same workflow a month ago.
Token budget engineering is becoming a primary concern quickly. The subsidized token era may be shorter lived than I hoped but competition is good and luckily workflows are adaptable.
It may be a mistake to "standardize" on one platform or another right now - I think we're in early innings here.
I've tried local models but at least what I've been able to experiment with so far is nowhere close. Looking for tips here too though!
Related insights
Codex Is Becoming My Daily Driver
After a year with Claude Code as my daily driver, the Codex desktop app's interface for managing many agent threads and automations has me reaching for it more and more.
I Built a Chrome Extension in 5 Minutes Instead of Searching for One
Things that used to be hard - like learning to build a Chrome extension - are now a few minutes away. So I built TabGrabber instead of hunting the web store for something I could trust.
You Don't Need Opus for Everything
Right model, right task. In Claude Code you can pin a model per skill—I drop to Sonnet or Haiku when the full power of Opus isn't needed to stay inside my limits.
Get More Like This
Follow along as I build and share what I learn
Found this helpful? Share it with your network!