laurels-marts

OpenAI has been a disaster lately.

gpt-6.1-sol was released 7 days after gpt-6-sol only because they were bleeding to fable-5.1/opus-5.5 from Anthropic. It was a desperation move and certainly ate into their margins massively. Since their margins on gpt-6.1-sol are much slimmer than on gpt-6-sol they have to use it somehow to preserve compute and make profit. Frankly I think OpenAI is a mess and is playing catch-up with Anthropic… I don’t know how they will recover.

100% in agreement. I pay for OAI sub and also use Codex exclusively at work for the past 8 months.

I tried out fable 5.1 the day it was released and coming from gpt-5.6-sol I was truly mind blown (both in terms of code and prose it was generating - outputs I could finally enjoy reading and looking at).

Then when opus 5.5 came out, again same thing + far cheaper and faster.

I went from using OAI exclusively the entire year to a point now where i haven’t touched one of their models in at least a few weeks now.

I think OAI has lost the plot. OAI models simplify have no taste. And I don’t mean in front-end design way (although that too). They have no taste in how the model writes code, how it writes prose, how it writes in-line comments, how it writes documentation, or how it even picks variable names. There’s just no taste throughout.

Anthropic models are very thoughtful and have so much taste all around.

> which shadowy government

A bit of false equivalency there

Why not use Codex CLI from inside WSL2? That’s that I do. I have Ubuntu LTS and while the VS Code extension works fine (for the most part) the CLI is where it’s at. The same goes for CC.

I actually love fable 5.1 prose. Before I was always on GPT models and I despised the writing. Actually I despised the writing they would do in the markdown docs, the writing in the responses was good. But there was always this weird thing that in the responses the model would write very clear explanations. Then if you ask it to document it what it ends up writing in the .md files is way too verbose, robotically lists facts and omits reasoning and gotchas. Too much unnecessary detail and too little detail it actually needed to capture. I absolutely hated GPT writing. Fable 5.1 though has the same concise, dense but logical and clear prose in both the responses and the documentation and the in-line code comments (oh yea Sol never wrote in-line code comments; and most of the time it deleted existing comments of code it didn’t even touch).

So yea, I find fable 5.1 writing to be excellent everywhere. I still use Sol daily though, but for things like config, quick research, fixes, code review etc. Feature work and writing is for fable 5.1.

That would have been great. Coulda gotten rid of an entire symlink folder structure.

So why should anyone slow down again? Because its like saying “i love you”?

Why are all these pro-regulation arguments so nonsensical…

> 60% of college/university graduates are women, but it’s not “necessarily a positive story for women overall,” since it also means women are disproportionately burdened with student-loan debt.

Sol is an absolute machine. I stopped doing parallel worktrees just because the cost of context switch outweighs the cost of waiting Sol to just finish the task it’s working on which is usually anywhere from 1-10mins.

I also like Codex CLI more than the Codex App bc it’s more scriptable and displays all the tool calls and reasoning whereas in the App it’s kind of folded away/obscured. This way as soon as I see a tool call fail (eg it tries to use jq assuming it’s available but it wasn’t so I take a note to set it up as it’s obviously useful for the agent to wrangle json).

I think its amazing what OpenAI have been able to squeeze out from a model like Sol thats much smaller in size than Fable.

Codex CLI has support for keybindings tho last time I checked they were quite limited and you cannot replicate full universal text editing experience.

Those that have, more will be given. Those that do not, more will be taken.

Refreshing that the author doesn’t try to conceal his character or thoughts.

The contents of his character and thoughts are quite objectionable though.

No one cares enough of zig to do it, people will just continue moving to Rust and similar.

You seem to think you know other people’s best interests better than they do.

Damn this is exciting. I love that gpt models are much faster, efficient and cheaper than Claude models. They are so fast even on high/xhigh that I don’t find myself using the parallel agent setup anymore much since its cognitively less demanding to just follow along what the model is doing and most tasks it will complete in <5-<10mins anyway.

I love that Codex is open source and issues like these can surface/be addressed publicly.

Wait what. I never used CC but use Codex CLI with 5.5 daily and authenticating has never been an issue. I even rolled skills that instruct it how to retrieve test user credentials for auth purposes.

Today using the devtools I asked it to reverse engineer the login auth flow of another app in our company and it created a nice browser-like headless script (with cookie jars etc) that emulates the entire Auth0 flow with all the internal API calls, redirect loops etc so that given username/password I end up with a valid JWT without having to open an actual browser instance and go through the login steps manually. Zero hesitation or questions asked.

I think this is in-line with OpenAI's philosophy. They see Codex agents as just tools for developer to use. They don’t try to imbibe them with “feelings”, “constitution” or “morality” the way Anthropic does.

And much faster!

My company gave me cursor license after I had already been using Codex CLI for months and VS Code for a decade.

I had absolutely no interest in their VS Code fork. The Agent Window was okay but buggy (eg wouldn’t load branches on Ubuntu via WSL2).

Overall used it a couple of times but still use Codex CLI as my main driver. Might try CC in the future esp. if they unban Fable.

i have shell scripts that create/tear down worktrees. the shell script finds unique unused ports across all existing worktrees and assigns in local .env upon worktree creation. when the worktree gets merged and is torn down the ports get released. secrets that are not worktree-specific i don't keep in local .env i inject via shell.

honestly creating these local scripts for automating the dev work was trivial and they combine well with all other cli tooling. thats why i havent tried any of the GUI apps yet. im not sure they're able to compete with my custom local setup that works exactly the way i want.

Are you complaining that government is not corrupt enough?

Wait so when u navigate to a different page/tab the url wouldn’t update?

You have been authoring HTML by hand for decades. Not every SWE is a FE dev.

For me it never updates and shutdown. It always just updates and restarts.

What I’ve heard often is that the customer might want to build a dashboard using AI on the fly.

Like imagine you have a site and there’s blank canvas. It has access to some data in the background. The user might be like “build a dashboard showing YoY performance of X and break down the shipments in a table by Y and Z”. Then the agent builds it and persists it such that when they log out and log back in they can see their custom dashboard they assembled themselves.

I’m basically doing lvl 3. There’s not a single port in my local worktree’s .env that’s not guaranteed to be unique across all worktrees. Skills for agent to start their own managed dev server, launch their own isolated instance of chrome etc. literally end-to-end code and debug the entire app. I do have to say though you have to know the app inside out and have to have a pretty well groomed backlog in order to run them all in parallel and actually benefit from it.

I’m a bit confused by this.

I migrated my Remix 2 app to react router 7 and I think it’s great.

Sure I have to pull in Express for server, Typescript for types, Vite for bundling etc but it’s not a lot of overhead. React Router mostly solves what I expect a modern framework to solve (also love v8 middlewares)… so what is Remix 3 adding? Are they really proposing that they will create a better bundler than Vite, a better UI renderer than React, a better types system than TypeScript? And for what? Such that you can say that you can build a “modern” app without any other deps other than Remix itself in your project?

Unless I misunderstand something but why not just roll a couple of helper shell scripts that create a new git worktree for you, copy over a local .env file (or any other config file) and populate it with ports/variables that are unique across all worktrees (to avoid collisions if you're on localhost, but you could also do this with Docker) and a helper script for tearing down a worktree after you've merged the changes back into main. I'm even supplying unique chrome debug ports and unique temp user data dirs for isolated automated testing in chrome via chrome devtools mcp. Like, I'm not sure why an external library/tool is needed for this.

Yes and the official docs even mention that if you’re on Windows you should run Codex CLI via WSL. Meaning, it’s specifically designed for unix systems.