← All episodes

Claude Opus 4.6 vs GPT-5.3 Codex: How I shipped 93,000 lines of code in 5 days

| 8 products mentioned
How I AI How I AI host
Watch on YouTube ai coding models claude opus vs gpt code generation ai-assisted development software engineering workflow design optimization architectural code review

Claire Vo tests Claude Opus 4.6 and GPT-5.3 Codex by using them to redesign her marketing website and refactor core application components, shipping 93,000 lines of code in 5 days. She finds that Opus 4.6 excels at creative, greenfield work and design, while GPT-5.3 Codex is superior for code review, architectural feedback, and catching edge cases—making them complementary tools in an AI engineering workflow.

Key takeaways
  • GPT-5.3 Codex models are too literal and struggle with creative, broad tasks; they excel instead at architectural review and identifying bugs, making them ideal for code quality assurance rather than building new features.
  • Claude Opus 4.6 is outstanding at generative greenfield work, planning long-running tasks, and producing high-quality design and UX, making it better suited for feature development and creative problem-solving.
  • The optimal workflow pairs Opus 4.6 for building features with Codex for rigorous code review, creating a principal engineer + product engineer dynamic that significantly accelerates development.
  • Codeex (OpenAI's desktop app) introduces useful features like Git primitives, skills as first-class citizens, and automations, but the conversational harness feels less mature than alternative IDEs for collaborative AI development.
  • Cursor remains the superior IDE harness for working with multiple models, offering better planning modes, exploration tools, and integrated code review via Bugbot that improve overall development velocity.
  • Opus 4.6 Fast costs 6x more than standard Opus but isn't always necessary; carefully match model selection to task complexity and budget to avoid unexpected token spending.

Recommendations (5)

GPT-4.5 Codex

"the new model GPT 5.3 Codex. Try saying that five times fast."

How I AI · ▶ 0:22

Claude Opus 4.6

"And Anthropic released their response, Opus 4.6 and Opus 4.6 Fast."

How I AI · ▶ 0:28

GitHub Desktop recommends

"I have always said you should invest in the GitHub desktop experience. It is a version of this. It's what I use all the time to manage my work across branches and across files."

How I AI · ▶ 6:43

Cursor
Cursor uses

"I have to give kudos to Cursor. I'm still a Cursor girl. I think Cursor does a good job of building harnesses for all of these models."

How I AI · ▶ 17:30

Devon
Devon uses

"I asked Devon today, how much code have I merged into GitHub in the last five days?"

How I AI · ▶ 21:38

Mentioned (3)

Hex
Hex "I like a more content dense site like Hex. Hex, you have a lovely site. I think you did a really ..." ▶ 14:24
Linear
Linear "So now from our chat, you can look at what's happening in GitHub, you can look at what's happenin..." ▶ 22:50
Granola
Granola "you can look at what's happening in GitHub, you can look at what's happening in Linear, you can l..." ▶ 22:54