Skip to main content

A New Wave of Vertically Integrated Coding CLIs



Yesterday, OpenAI released Codex, a response to Anthropic’s Claude Code. Here’s a summary from ChatGPT what they actually do: 

Anthropic’s Claude Code

On February 24, 2025, Anthropic quietly rolled out Claude Code, a command‐line interface built atop its Claude 3.7 Sonnet model. Designed to “live in your terminal,” it ingests your codebase context and lets you invoke natural‐language commands—everything from scaffolding new modules to refactoring existing functions—without leaving the command line. By embedding its own LLM at the core of the developer workflow, Anthropic ensures feature parity between the model’s evolving capabilities and the tool’s UX.

OpenAI’s Codex CLI

Just weeks later, on April 16 and 17, 2025, OpenAI published its open‐source Codex CLI, bringing the latest OpenAI reasoning engines straight to the shell. Out of the box, it supports multimodal inputs—text, screenshots, even hand‐drawn diagrams—and provides a granular approval workflow (suggest, auto‐edit, full‐auto). Like Claude Code, it requires zero server setup and rides on the same releases cycle as OpenAI’s flagship models, promising instant access to every new improvement in GPT 5’s reasoning, code understanding, and context handling prowess.

I’ve provided ChatGPT with a couple of Model agnostic contenders, and I’ve added more context where I saw it fit: 

Model Agnostic Contenders

In parallel, a host of model agnostic tools continue to flourish:

  • Roo Code: A VS Code extension and CLI agent that connects to “any OpenAI compatible or custom API,” offering dedicated modes for coding, architecture planning, and debugging. This is actually a fork of Cline (below)
  • Cline: An LLM assistant inside VS Code, interoperable with Claude, OpenAI, Google Gemini, and OpenRouter, thanks to its adherence to the Model Context Protocol. This was origianlly released as open source by Anthropic.
  • Windsurf (formerly Codeium): An intelligent IDE overlay capable of multi file edits and project wide analysis, powered by its own AI engine stack.
  • Cursor: A VS Code fork by Anysphere Inc., featuring AI driven autocompletion, generation, and codebase querying via a proprietary agent.

And there are several others in this space. 

I further wanted to see what ChatGPT can come up with in terms of Pros and Cons: 

Vertical Integration: Pros and Cons

The rapid succession of Claude Code and Codex CLI highlights a broader strategy: vertical integration of tooling and models.

Pros:

  • Optimized performance and tighter security by keeping code and model inference localized.
  • Seamless updates, with every model enhancement immediately reflected in the CLI.
  • Unified support, reducing configuration headaches and third party dependencies.

Cons:

  • Vendor lock in, making it harder to switch providers without rewriting workflows.
  • Stifled innovation in the tooling ecosystem, as smaller players compete against deep pocketed giants.
  • Regulatory scrutiny, since consolidation in developer tooling may raise antitrust flags if major LLM houses begin acquiring independent CLI startups.

It think it’s a pretty good summary and is essentially generic pros and cons in verticalization business models.  

Conclusions

I tried to keep this intentionally short, however, I’m writing the conclusions without AI:

I’ve been playing around with Codex for not more than 30 minutes today, but I can already say that it’s a worthy competition to Claude Code. It’s slower but possibly better thanks in part to the new o4-mini model that was released alongside. In my 30 minute experiment, I asked it to refactor a component, and add some functionality from another component so that we can get rid of that other component. It did it on the first try but it took a good 5 minutes. It still lacks some of the polish from Claude Code but OpenAi has the resources to make this better. Also, it doesn’t give me a cost analysis the way Claude Code does – so I’m basially in the dark how much the 5 minutes cost me and I have check on OpenAi’s API website (it was 61 cents – in-line with what I would have paid with Claude Code just based on experience). 

Fundamentally, it’s exciting to see vertical integration by the large LLMs and it gives a bit of weight to the statement I made in my Vibe-coding post (URL), namely that I think we’ll see more specialization in the models and with that, more specialized tools that follow. I can’t say if this is a positive development overall, but it’s certainly exciting. 


Labels:

Comments

Post a Comment

Popular posts from this blog

The case for central bank digital currency as public infrastructure to enable digital assets

I have dabbled a fair amount with all sorts of crypto currencies and their respective permissionless networks. In fact, I have been dabbling since 2012 which is by my last count a whopping 12 years. While I have always maintained that I do believe the general concept for digitization and programmability of assets is on the right path, its implementation, the user experience, the accessibility, the fraudulent activities, and the overall inefficiencies permissionless DLTs have, never made me into a true believer. I have stated that opinion on several occasions, here , here and here . There are still barriers to entry when it comes to digitization of assets: sustainable- and interoperable infrastructure. To illustrate this, I recently asked a notary public here in Zurich, why they can’t store the notarized documents as PDFs, the answer surprised me: because they must keep records for at least 70 years. Now, think about what would have happened if we stored these documents on floppy disks...
Post a Comment
Read more

Apples Vision Pro Headset strategy is all about its Arm-chips.

  Apple has given us a vision of what their VR and AR future might entail. But as have others pointed out numerous times, the whole point of the showcase at the WWDC 23 was to let people experiment, I’ve heard others say that it’s like the launch of the Apple Watch when Apple didn’t really know what would become of it. This is similar and yet different.  Just like the Apple Watch (and the iPad before it), Apple sought to porting its whole ecosystem onto a watch – granted, the Apple Watch can’t live on its own and a better comparison would probably be the iPad. The iPad can live without any other Apple device and unlike the iPhone, never really had a clearly defined function other than to watch movies and browse the web. It was not until it gained the ability to be used with a pencil that artists and designers started to explore the potential.  I’m trying to point out that Apple took 5 years from the first iPad in 2010 to the iPad Pro with pencil in 2015 to find its “kille...
Post a Comment
Read more

Will smart phone cameras be better than digital mirrorless cameras?

  If you believe Terushi Shimizu or rather, the way the press is formulating it , then camera phones will have better image quality in 2024 than your trusty DSLR or mirrorless digital camera. He backs this up with sensor technology advancements and computational photography. He has a point.     However, as a digital camera enthusiast myself, I must strongly disagree with this point of view. The message might be interpreted in such way that its meaning reflects a view that we are no longer bound by physics to get the best image quality.     The thing is this, the bigger your camera sensor, the more photons it can capture. However, this comes at the realization that big sensors require big lenses which in turn makes the camera big and heavy. I’m simplifying of course, but that’s physics. For camera makers it is therefore always a question of tradeoffs: do you want better image quality or do you want a smaller and lighter camera. Camera phones or cameras in smartph...
Post a Comment
Read more