/Tech12h ago

OpenClaw steward Peter Steinberger and academic Jason Lee prefer OpenAI's Codex over Claude Code, citing GPT-5.5 performance

Some developers still use Claude Opus for frontend tasks.

2432K64559592.2K

#111

Original post

Peter Yang@petergyang

I used to be a die-hard Claude Code user.

Codex has won me over because:

→ GPT-5.5 is excellent → Fast mode + generous limits = more reps → Little touches like steering, auto remote control on phone, etc

But most of all Codex's browser and computer use capabilities are simply goated. I built so many workflows relying on those two things alone instead of hunting for APIs.

I still use Claude Code too. The app seems to be getting better and the design and frontend capability of Opus is still much better than GPT. Whenever Fable comes back that's another reason to go back.

Honestly, I hope these two compete forever and other players (Cursor/Grok, Gemini, etc) all stay competitive.

This way the builder keeps winning 🙂

8:32 PM · Jun 19, 2026 · 590.2K Views

Sentiment

Positive users highlight Codex surpassing Claude in browser and computer use tasks as the reason it has won them over from prior loyalty, while some criticize its weaker UI design performance.

Pos

85.7%

Neg

14.3%

35 comments with sentiment.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Posts from X

Most Activity

VIEWS31KBOOKMARKS17

Peter Yang@petergyang

Will make a video on Codex browser use use cases soon via my channel, subscribe here: https://www.youtube.com/@PeterYangYT?subscribe

1d31K2917

LIKES42

Matthew Berman@MatthewBerman

@petergyang Computer use codex is so good

2d3.6K421

REPLIES4

Anthony Morris ツ@amorriscode

@petergyang happy to take any feedback for the CC desktop app 🙏

1d1.7K41

Troy@OderintMetuant

@petergyang The integration with ChatGpt Images 2.0 is another big advantage of Codex.

I can insert great art directly into my files, webpages, etc from Codex with no additional setup.

2d11.4K331

Will Cook@cookie1234593

@petergyang Using both obsessively gets you something like this

2d1.3K3

Jason Lee@jasondeanlee

Agree. Claude code math is bad, computer use is bad and costs $$

Peter Yang@petergyang

I used to be a die-hard Claude Code user.

Codex has won me over because:

→ GPT-5.5 is excellent → Fast mode + generous limits = more reps → Little touches like steering, auto remote control on phone, etc

But most of all Codex's browser and computer use capabilities are simply goated. I built so many workflows relying on those two things alone instead of hunting for APIs.

Honestly, I hope these two compete forever and other players (Cursor/Grok, Gemini, etc) all stay competitive.

This way the builder keeps winning 🙂

6h3.3K190

Mark Huthmaker@SpartanDetermin

The “working backwards” approach to front end design has made the Codex app #1 for me, despite using claude code for over a year.

For a product that already has the backend built, I’ll fine-tune how I want the app to actually look via Nano Banana & Image2. Then I present the final image(s) to Codex app and it does the wiring + frontend work to get there. It takes a few iterations, and eats up a lot of tokens, but it works.

2d28012

Peter Yang@petergyang

@ktimesk Not at all the compaction is fine

2d4.3K12

Peter Yang@petergyang

@OderintMetuant Yes I think this will be a long-term competitive advantage for design. Most designs will have some illustrations or images tied up to them.

2d8K15

Kaushal Kashyap Ⓥ@ktimesk

@petergyang I don't understand. Can you expand on your response?

2d779

Another dev joined X@ahanotherdev

@petergyang To solve the problem. Check this out if you want to use Claude as subagents and codex as orchestrator https://github.com/jnopareboateng/codex-claude-subagents

1d5302

Kaushal Kashyap Ⓥ@ktimesk

@petergyang Doesn't the lack of 1M context bother you? For me, Codex starts compacting within 15-30 minutes pf my session.

2d5K4

Ben Schauerhamer@America_Dev

@petergyang Have you fully dived in with @Grok Build or Cursor?

I do agree that Codex is amazing with what it can do but with the updates that have come out with Grok Build over the last week, watch out.

6h491

netrunner@netrunner_btc

@petergyang opus frontend still wins for me, i run both but design work goes to claude every time

2d3042

Joey Smith@JoeySmith639425

@petergyang I agree, but for anyone reading this: You can use Claude models inside the Codex App with most native functions preserved. Codex Browser and Computer use are not available yet sadly, perhaps interesting for anyone who likes the Codex app but wants to use Claude models.

17h49911

Nelly;@nrqa__

@petergyang Codex winning you over for reps makes so much sense

1d25311

Carter@CarterKoWang

@petergyang the codex app is amazing. i’m surprised anthropic still hasn’t shipped a better ot at least comparable app for claude code by now

1d4123

P.B.@ruysbroek

@petergyang I am back to Codex too: it drifted a few time on iOS project with not clear understanding why, but remote control is too much of a convenience.

1d3743

Blaida@kedytcom

@petergyang Tools don’t matter.

Output does.

Use Codex for speed. Use Claude for taste. Ship faster than people debating both.

1d1603

Dirk Kok · AI agents@dirkkok

The smarter model is the worse employee. Opus is more capable, and that is exactly why it redesigns things I never asked it to. Codex is more literal, so I spend far less time undoing work I did not want. On a real codebase, predictable beats clever, and the longer limits just mean more reps at it

1d2761