Codex Just 10x’d Claude Code Projects
OpenAI released an official Codex plugin for Claude Code that allows developers to use GPT-4o for code reviews within their Claude workflows. The speaker tested both tools and found they complement each other well - Claude Code excels at planning and creative outputs while Codex is better at code reviews and catching edge cases.
Summary
OpenAI has released an official Codex plugin for Claude Code, making it easier for developers to incorporate GPT-4o into their existing Claude workflows for code reviews and additional oversight. While people were already combining these tools, this plugin streamlines the process and can be used for free with a ChatGPT subscription. The speaker analyzed benchmarks comparing Claude Opus 3.5 with GPT-4o, finding that while Opus leads slightly in one benchmark (SWE-bench verified), GPT-4o outperforms Opus across most other coding benchmarks by significant margins (10-13 points in some cases) while being more cost-effective. Through research on social platforms, the speaker identified complementary strengths: Claude Code tends to over-engineer, be token-hungry, and miss edge cases in long runs, while Codex struggles with planning, asking good questions, and creative outputs. This makes them natural complements - many users plan and build initial versions with Claude Code, then use Codex for execution, production deployment, and reviews. The speaker conducted a practical test by giving both tools identical prompts to build a dungeon crawler game. Codex took longer but produced a more polished, less pixelated result that felt more like a professional application. The speaker then used Codex's adversarial review feature to analyze the Claude-built game, which identified critical bugs including a soft-lock scenario where players could become permanently stuck and data loss issues. After implementing Codex's suggested fixes in the Claude-built game, the functionality improved significantly. The plugin offers various functions including standard reviews, adversarial reviews, and rescue operations, essentially acting like additional skills that can run in the background.
About this episode
Full courses + unlimited support: https://www.skool.com/ai-automation-society-plus/about All my FREE resources: https://www.skool.com/ai-automation-society/about Apply for my YT podcast: https://podcast.nateherk.com/apply Work with me: https://uppitai.com/ My Tools💻 14 day FREE n8n trial: https://n8n.partnerlinks.io/22crlu8afq5r Code NATEHERK to Self-Host Claude Code for 10% off (annual plan): https://www.hostinger.com/vps/claude-code-hosting Voice to text: https://ref.wisprflow.ai/nateherk X Article: https://x.com/reach_vb/status/2038670509768839458 OpenAI just released an official Codex plugin for Claude Code, and it's a surprisingly strong combo. In this video I break down the benchmarks between Opus 4.6 and GPT 5.4, share what the community has been saying about the strengths and weaknesses of each tool, and then put them head to head with a live game build and an adversarial code review. If you're using Claude Code and haven't tried bringing Codex into your workflow yet, this will show you exactly why you should. Sponsorship Inquiries: 📧 [email protected] TIMESTAMPS 0:00 What Is the Codex Plugin 1:09 Opus 4.6 vs GPT 5.4 Benchmarks 2:09 Strengths & Weaknesses of Each 3:18 Using Both Tools Together 3:37 How to Set It Up 4:35 Live Adversarial Review Demo 6:56 Head-to-Head Game Build 9:53 Why Not Just Use Codex? 10:39 Feeding Codex Review Back to Opus 12:48 Final Thoughts
Key Insights
- GPT-4o outperforms Claude Opus 3.5 across most coding benchmarks by significant margins (10-13 points in some cases) while being more cost-effective than Opus
- Claude Code's weaknesses include over-engineering, being token-hungry, and missing edge cases in long runs, while Codex struggles with planning, asking good questions, and creative outputs
- Many developers use a hybrid approach where they plan and build initial versions with Claude Code, then bring in Codex to execute the rest and push to production
- In a direct comparison building the same dungeon crawler game, Codex produced a more polished, less pixelated result that felt more like a professional application, despite taking longer to complete
- Codex's adversarial review identified critical bugs in the Claude-built game including a soft-lock scenario where players could become permanently stuck and data loss issues
Topics
Transcript
[0:00] So today, OpenAI released an official codec plugin for cloud code. You can see here it says if you already use cloud code, this codeex plugin gives you a simple way to pull codeex into that same workflow. Now this isn't like groundbreaking technology. People have already been kind of using a combination of codec and claude code together, but this plugin makes it a lot easier. And so obviously I've been focusing a lot on cloud code, but I have played around with codecs a little bit and they felt very similar to me. But as of lately, I've been seeing more and more people talk about how they've been using codecs inside of their cloud code workflows…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Nate Herk | AI Automation
Fable 5 + Karpathy’s LLM Wiki is Basically Cheating
The video demonstrates how to build a personal LLM-powered knowledge base using Obsidian and Claude, inspired by Andrej Karpathy's approach. By ingesting various data sources (YouTube transcripts, PDFs, URLs) into an interconnected wiki structure, users can create a 'second brain' that helps AI agents understand context and relationships across knowledge domains.
How Claude is Creating a New Generation of Millionaires
Claude, an AI built by Anthropic, is enabling non-technical founders to build million-dollar software companies by automating development work. The transcript outlines why Claude is outpacing competitors, provides real examples like Vulcan, and offers a four-step framework for anyone to start building with Claude today.
How Anthropic Engineers Actually Prompt Fable 5
A guide to effectively prompting Claude Fable 5, covering six key techniques including providing context/intent, using negative prompting, letting the model act once ready, verifying outputs, avoiding reasoning requests, and keeping instructions concise. The speaker emphasizes that Fable 5 is expensive and should only be used 5-15% of the time, with pricing at $10 per million input tokens and $50 per million output tokens.
Stanford's Method Turns Claude Into a PHD Level Research Team
A researcher demonstrates the STORM method from Stanford, which uses five expert perspectives (practitioner, academic, skeptic, economist, historian) to create verified research reports. The method produces 25% more organized articles than competing approaches and is packaged as a reusable Claude skill that generates HTML briefings with peer-reviewed citations.
Is Claude Mythos Coming?
A YouTuber analyzes the brief appearance of 'Claude Mythos' on Anthropic's API, arguing it signals a marketing move rather than an imminent public launch. Despite competitive pressure from OpenAI and IPO timing, the creator believes Mythos capabilities will quietly fold into future Opus models rather than release publicly under that name. Three possible scenarios are outlined ranging from a limited gated release to Mythos remaining permanently restricted to vetted security partners.