OpenAI’s Windsurf deal is off, and Windsurf’s CEO is going to Google

Yes, my CC usage is regularly $50-$100 per day, so their Max plan is absolutely great value that I don’t expect to last.

Can you give me an idea of how much interaction would be $50-$100 per day? Like are you pretty constantly in a back and forth with CC? And if you wouldn’t mind, any chance you can give me an idea of productivity gains pre/post LLM?

Pretty easy to hit $100 an hour using Opus on API credits. The model providers are heavily subsidized, the datacenters appear to be too. If you look at the Coreweave stuff and the private datacenters it starts looking like the telecom bubble. Even Meta is looking to finance datacenter expansion - https://www.reuters.com/business/meta-seeks-29-billion-priva...

The reason they are talking about building new nuclear power plants in the US isn't just for a few training runs, its for inference. At scale the AI tools are going to be extremely expensive.

Also note China produces twice as much electricity as the United States. Software development and agent demand is going to be competitive across industries. You may think, oh I can just use a few hours of this a day and I got a week of work done (happens to me some days), but you are going to end up needing to match what your competitors are doing - not what you got comfortable with. This is the recurring trap of new technology (no capitalism required.)

There is a danger to independent developers becoming reliant on models. $100-$200 is a customer acquisition cost giveaway. The state of the art models probably will end up costing hourly what a human developer costs. There is also the speed and batching part. How willing is the developer to, for example, get 50% off but maybe wait twice as long for the output. Hopefully the good dev models end up only costing $1000-$2000 a month in a year. At least that will be more accessible.

Somewhere in the future these good models will run on device and just cost the price of your hardware. Will it be the AGI models? We will find out.

I wonder how this comment will age, will look back at it in 5 or 10 years.

Re productivity gains, CC allows me to code during my commute time. Even on a crowded bus/train I can get real work done just with my phone.

Yes, a lot of usage, I’d guess top 10% among my peers. I do 6-10hrs of constant iterating across mid-size codebases of 750k tokens. CC is set to use Opus by default, which further drives up costs.

Estimating productivity gains is a flame war I don’t want to start, but as a signal: if the CC Max plan goes up 10x in price, I’m still keeping my subscription.

I maintain top-tier subscription to every frontier service (~$1k/mo) and throughout the week spend multiple hours with each of Cursor, Amp, Augment, Windsurf, Codex CLI, Gemini CLI, but keep on defaulting to Claude Code.

Unless you're getting paid for your commute, you're just giving your employer free productivity. I would recommend doing literally anything else with that time. Read a book, maybe.

What's your workflow if I may ask? I've been interested in the idea as well.

I am curious what kind of code development you are doing with so many subscriptions?

Are you doing front end backend full stack or model development itself?

Are you destilling models for training your own?

I have never heard someone using so much subscription?

Is this for your full time job or startup?

Why not use qwen or deep seek and host it yourself?

I am impressed with what you are doing.

Thank you for your perspective. I’ve been staring at Claude Code for a bit and I think I will just pull the trigger.

It's for a paid side gig.

The project is just a web backend. I give Claude Code grunt work tasks. Things like "make X operation also return Y data" or "create Z new model + CRUD operations". Also asking it to implement well-known patterns like denouncing or caching for an existing operation works well.

My app builds and runs fine on Termux, so my CLAUDE.md says to always run unit tests after making changes. So I punch in a request, close my phone for a bit, then check back later and review the diff. Usually takes one or two follow-up asks to get right, but since it always builds and passes tests, I never get complete garbage back.

There are some tasks that I never give it. Most of that is just intuition. Anything I need to understand deeply or care about the implementation of I do myself. And the app was originally hand-built by me, which I think is important - I would not trust CC to design the entire thing from scratch. It's much easier to review changes when you understand the overall architecture deeply.

Is there a cheap version for hobbyists? Or what’s the best thing for hobbyists to use, just cut and paste?

I am curious what kind of development you’re doing and where your projects fall on the fast iteration<->correctness curve (no judgment). I’ve used CC Pro for a few weeks now and I will keep it, it’s fantastically useful for some things, but it has wasted more of my time than it saved when I’ve experimented with giving it harder tasks.

Cursor at 20$/M is pretty great

I've been enjoying Zed lately

Claude Code pro is ~$20USD/ month and is nearly enough for someone like me who can’t use it at work and is just playing around with it after work. I’m loving it.

How do you use Claude Code via your phone?

Claude Code with a Claude subscription is the cheap version for current SOTA.

"Agentic" workflows burn through tokens like there's no tomorrow, and the new Opus model is so expensive per-token that the Max plan pays itself back in one or two days of moderate usage. When people reports their Claude Code sessions costing $100+ per day, I read that as the API price equivalent - it makes no sense to actually "pay as you go" with Claude right now.

This is arguably the cheapest option available on the market right now in terms of results per dollar, but only if you can afford the subscription itself. There's also time/value component here: on Max x5, it's quite easy to hit the usage limits of Opus (fortunately the limit is per 5 hours or so); Max x20 is only twice the price of Max x5 but gives you 4x more Opus; better model = less time spent fighting with and cleaning up after the AI. It's expensive to be poor, unfortunately.

> Pretty easy to hit $100 an hour

I don’t see how that can be true, but if it is…

Either you, or I are definitely use Claude Code incorrectly.

It’s a wild frontier, but as a recent convert to CC, I would say go for it.

It’s so stupid fast to get running that you aren’t out anything if you don’t like it.

There was no way I was going to switch to a different IDE.

I’m a founder/CTO of an enterprise SaaS, and I code everything from data modeling, to algos, backend integrations, frontend architecture, UI widgets, etc. All in TypeScript, which is perfectly suited to LLMs because we can fit the types and repo map into context without loading all code.

As to “why”: I’ve been coding for 25 years, and LLMs is the first technology that has a non-linear impact on my output. It’s simultaneously moronic and jaw-dropping. I’m good at what I do (eg, merged fixes into Node) and Claude/o3 regularly finds material edge cases in my code that I was confident in. Then they add a test case (as per our style), write a fix, and update docs/examples within two minutes.

I love coding and the art&craft of software development. I’ve written millions of lines of revenue generating code, and made millions doing it. If someone forced me to stop using LLMs in my production process, I’d quit on the spot.

Why not self host: open source models are a generation behind SOTA. R1 is just not in the same league as the pro commercial models.

The SOTA models will always run in data centers, because they have 5x or more VRAM and 10-100x the compute allowance. Plus, they can make good use of scaling w/ batch inference which is a huge power savings, and which a single developer machine doesn’t make full use of.

vibetunnel.sh perhaps

Why “no capitalism required”? Competition of this kind is only possible with capitalism.

> If someone forced me to stop using LLMs in my production process, I’d quit on the spot.

Yup 100% agree. I’d rather try to convince them of the benefits than go back to what feels like an unnecessarily inefficient process of writing all code by hand again.

And I’ve got 25+ years of solid coding experience. Never going back.

When you say generation behind, can you give a sense of what that means in functionality per your current use? Slower/lower quality, it would take more iterations to get what you want?

Zed is fantastic. Just dipping my toes in agentic AI, but I was able to fix a failing test I spent maybe 15 minutes trying to untangle in a couple minutes with Zed. (It did proceed to break other tests in that file though, but I quickly reverted that.)

It is also BYOA or you can buy a subscription from Zed themselves and help them out. I currently use it with my free Copilot+ subscription (GitHub hands it out to pretty much any free/open source dev).

you can easily reach 50$ per day. by force switching model to opus /model opus it will continue to use opus eventhough there is a warning about approaching limit.

i found opus is significantly more capable in coding than sonnet, especcially for the task that is poorly defined, thinking mode can fulfill alot of missing detail and you just need to edit a little before let it code.

Everywhere I've worked as a programmer you're just paid to do your job. If you get some of it done on your commute what difference does it make?

It's interesting to work with a number of people using various models and interaction modes in slightly different capacities. I can see where the huge productivity gains are and can feel them, but the same is true for the opposite. I'm pretty sure I lost a full day or more trying to track down a build error because it was relatively trivial fpr someone to ask CC or something to refactor a ton of files, which it seems to have done a bit too eagerly. On the other hand, that refactor would have been super tedious, so maybe worth it?

Not really, it's possible with any market economy, even a hypothetical socialist one (that is, one where all market actors are worker-owned co-ops).

And, since there is no global super-state, the world economy is a market economy, so even if every state were a state-owned planned economy, North Korea style, still there would exist this type of competition between states.

I mean, if you wanna get technical, many companies in Silicon Valley are worker-owned (equity compensation)

They are not worker owned, they have some small amount of worker ownership. But the majority of stock is never owned by workers, other than the CEO.

It’s definitely easy with an API key I hit 200$ in an evening. I didn’t think that could be possible. Horrifying

Consider also that VC funds often have pension funds as their limited partners. Workers have a claim to their pension, and thus a claim to the startup returns that the VC invests in.

So yeah it basically comes down to your definition of "worker-owned". What fraction of worker ownership is necessary? Do C-level execs count as workers? Can it be "worker-owned" if the "workers" are people working elsewhere?

Beyond the "worker-owned" terminology, why is this distinction supposed to matter exactly? Supposing there was an SV startup that was relatively generous with equity compensation, so over 50% of equity is owned by non-C-level employees. What would you expect to change, if anything, if that threshold was passed?

> data modeling, to algos, backend integrations, frontend architecture, UI widgets, etc. All in TypeScript, which is perfectly suited to LLMs because we can fit the types and repo map into context without loading all code.

Which frameworks & libraries have you found work well in this (agentic) context? I feel much of the js lib. landscape does not do enough to enforce an easily-understood project structure that would "constrain" the architecture and force modularity. (I might have this bias from my many years of work with Rails that is highly opinionated in this regard).

To be clear, this is a lot of full-scale reading and (re)writing, without any rules, promots, "agents"/code to limit your resource usage, right?

Nobody's asking for $200 in single-line diffs in less than a day - right?

It’s not about a single line diff but the same prompts with cursor does not end costing that much

wow. haven't tried Opus but Sonnet 4 is already damn good.

Have you been human before? competition for resources and status is an instinctive trait.

It rears its head regardless of what sociopolitical environment you place us in.

You’re either competing to offer better products or services to customers…or you’re competing for your position in the breadline or politburo via black markets.

    > I’ve written millions of lines of revenue generating code

This is a wild claim.

Approx 250 working days in a year. 25 years coding. Just one million lines would be phenom output, at 160 lines per day forever. Now you are claiming multiple millions? Come on.

>less time spent fighting with and cleaning up after the AI.

I've yet to use anything but copilot in vscode, which is 1/2 the time helpful, and 1/2 wasting my time. For me it's almost break-even, if I don't count the frustration it causes.

I've been reading all these AI-related comment sections and none of it is convincing me there is really anything better out there. AI seems like break-even at best, but usually it's just "fighting with and cleaning up after the AI", and I'm really not interested in doing any of that. I was a lot happier when I wasn't constantly being shown bad code that I need to read and decide about, when I'm perfectly capable of writing the code myself without the hasle of AI getting in my way.

AI burnout is probably already a thing, and I'm close to that point already. I do not have hope that it will get much better than it is, as the core of the tech is essentially just a guessing game.

100-200 lines per day, written, debugged, tested and deployed, is normal performance, isn't it? I think I could do it if worked for 8 hours.

Unfortunately it's called war and it appears to be part of human nature.

Right, this is having Claude Code just running as an agent doing a lot of stuff. Also tool use is a big context hog here.

Uh... Totaling +1000 at the end of a work week is an easy thing to do, especially if working on a new/evolving product.

Your excellent comments make me grateful that I am retired and just work part time on my own research and learning. I believe you when you say professional developers will need large inference compute budgets.

Probably because I am an old man, but I don’t personally vibe with full time AI assistant use, rather I will use the best models available for brief periods on specific problems.

Ironically, when I do use the best models available to me it is almost always to work on making weaker and smaller working models running on Ollama more effective for my interests.

BTW, I have used neural network tech in production since 1985, and I am thrilled by the rate of progress, het worried about such externalities as energy use, environmental factors, and hurting the job market for many young people.