There is a world of difference of having tens of millions of dollars and paying hundreds of dollars a month for AI subscriptions vs someone making 100k / year doing the same.
For the DHH's of the world that's not even something to think about, but for most people who are working (even in tech), now we're talking 2-8%+ of your income depending how much you use AI. Just to have the "luxury" of being able to experiment with it. I use air quotes on luxury because I personally find a lot of AI results to be quite bad, but then again, I'm not spending hundreds of dollars a month on the premium models.
I make 55k working at a dental office and a restaurant and I'm paying 200$ a month on open source models because I'm having an absolute blast with them.
Plenty of jobs have expenses built into them. Maybe a subscription is more than it used to be, I had great jobs on a used 200$ thinkpad, but it's still cheaper than being a photographer. And 100k is more than most jobs.
Software dev, even if it's all agents now, is still pretty dang economical
It's easier to enjoy it if you don't see how the sausage is made.
I am definitely not having a blast at work with LLMs. Everyone in the team has become simple rubber-stampers for whatever the AI puts out. There are mountains of documentation which gives a false sense of understanding of how things work, but the code repository itself could have been written in Brainfuck and it would make almost no difference. And we keep getting told by management that this is fine. We've become simple janitors in the slop hall.
Everyone in the team has become simple
rubber-stampers for whatever the AI puts out.
Whoever figures out better ways to:
1. Present a big AI PR to a fellow human for review
2. Understand a big AI PR, as a human reviewer
...is going to be a hero. This is one of the biggest unsolved problems of this new era.
I've been experimenting with this, but it's early days. My efforts are mostly along the lines of having AI generate artifacts that guide the reviewer through the PR in execution flow order, explaining the reason for changes along the way.
I don't think that's the way forward. We don't accept mega-PRs from people, why accept them from machines? The best practice of small, focused commits - whether it be from a person or a LLM - will win out in time once we're past this manic phase.
Have you tried OpenRouter and the DeepSeek-like models? If you try to optimize a little you could probably slash those costs.
I send a ton of requests and even reaching $10 for the month would take more calls than I have capacity to review and use. And as I said, that's for a ton of requests.
20 a month and using sonnet mostly for random fun stuff.
At work, it seems a bit different - I'm certainly having to declaudify things it can be hard starting there's a lot to dig into - you can see it's overwritten but need to understand every detail to sort it out.
You can pay $5 for deepseek 4.1 flash and be on the pareto frontier. And you can pay between $20-200/month for a huge amount of usage of the absolute frontier models. I was concerned about costs a year ago but with the proliferation of great cheap models, I think it's no longer a concern at all. I'd be fine if I could only use 4.1 flash.
I think the fear, threat, and valuation, of the frontier labs is that they will cross some line and have models that are infinitely smarter and more capable than everyone else. If that should happen, your deepseek account would be meaningless.
But I also think that premise is bullshit and not gonna happen.
You think 2-8% is a problem? There's a large world outside the US. "Hundreds of dollars" per month is most of my salary, I can afford that only by going hungry half the time.
Probably going to if I want to keep my job, since I'll be making about the same amount of money moving boxes or cleaning toilets, only in much less comfortable conditions.
It's essentially trying to turn programming into a micro-transaction based pay to play / pay to win outcome, a concept that was gaming related for decades.
The premium models are worth 10% of your income at least.
There is no future without these models. They truly are an entirely different league than the cheap stuff. You can move mountains with Astra or Opus 5.5.
10% of a software engineer salary is a lot of money. That's just ridiculous.
Like the other commentor said, its just going to be a net negative, because whatever productivity increase you supposedly do get, you will see exactly 0 of that translated into salary increase.
Have you tried to negotiate a 10% pay raise because “you can move mountains with Astra or Opus 5.5”? I’d wager you’re more likely to get a “the fuck do we need you for then” than that pay raise.
The whole AI issue is and always has been a social question. It's hardly a ridiculous concept to have a skilled worker control a $10k budget to buy tools. In many fields that is normal.
Previously, the tools in our field were free or trivial cost. Seems like that has changed, that's OK, there are models to solve that.
I teach a couple courses currently, there is an enormous gap between what my students can create with AI and what I can create with AI. I don't think it's that much smaller than the difference of what we can create without, the only difference is that the schedules are compressed.
I swap between Fable and GLM 5.X and both work fine. Fable tends to do stuff that I didn't want it to do, and claude is probably a bit more capable as a harness (though I haven't spent the effort configuring pi properly yet so that's probably on me). Otherwise it's hard to spot the difference.
Good point. Most of us at the conference are still trying to make a living with Rails. That's not a concern for him. Can he remember what it was like to have to do that? Probably, but he feels very differently waking up each day than the rest of us.
Here is the description for the Rails World 2026 Opening Keynote:
"Opening Keynote highlighting what's new in Rails, what's coming next, and where Rails is headed in the future."
DHH literally did not cover that. Not even close!
The first response in the conference chat described it as the "worst opening keynote possible to setup the Rails World conference." While someone soon disagreed with that, another guy later said he felt it undermined the later talks, and someone else said the Laracon conference had more about what the framework is doing in the AI age. He barely mentioned Rails. I get how he was excited about his experiments but I certainly wish he had focused more on what Rails is doing for this new era.
