The macOS app version of Codex I have doesn't show reasoning summaries, just simply 'Thinking'.

Reasoning deltas add additional traffic, especially if running many subagents etc. So on large scale, those deltas maybe are just dropped somewhere.

Saying that, sometimes the GPT reasoning summary is funny to read, in particular when it's working through a large task.

Also, the summaries can reveal real issues with logic in prompts and tool descriptions+configuration, so it allowing debugging.

i.e. "User asked me to do X, system instructions say do Y, tool says Z which is different to what everyone else wants. I am rather confused here! Lets just assume..."

It has previously allowed me to adjust prompts, etc.

It's useful when using prism, and for exploratory research & code.

I do want to see as it allows me to course correct.

Yes, Artificial Goofy Intelligence

I'm switching because of the higher usage limits, 2x speed mode that isn't billed as extra usage, and much more stable and polished Mac app.

FWIW, I've found Codex with GPT-5.4 to be better than Opus-4.6; I would say it's at least worth checking out for your use case.

at least for our scope of work (data, interfacing with data, building things to extract data quickly and dump to warehouse, resuming) claude is performing night and day better than codex. we're still continuing tinkering with codex here to see if we're happy with it but it's taking a lot more human-in-the-loop to keep it from going down the wrong path and we're finding that we're constantly prompt-nudging it to the end result. for the most part after ~3 days we're not super happy with it. kinda feels like claude did last year idk. it's worth checking out and seeing if it's succeeding at the stuff you want it to do.

Wait for new GPT release this/next week and then decide based on benchmarks. That is what I will do.

One main thing is to de-couple the repos from specific agents e.g. use .mcp.json instead of "claude plugins", use AGENTS.md (and symlink to CLAUDE.md) and so on.

I love this because I have absolutely 0 loyalty to any of these companies and once Anthropic nerfs I just switch to OpenAI, then I can switch to Google and so on. Whichever works best.

Honestly, just try it. I used both and there's no reason to not try depending on which model is superior at a given point. I've found 5.4 to be better atm (subject to change any time) even though Claude Code had a slicker UI for awhile.