AI coding at home without going broke • Hacker News Board

Re: AI coding at home without going broke

#181

Post by alecco » Sat, Jun 13, 2026, 9:30 PM UTC

Yoric wrote: post #78

atreids wrote: post #2

I find just going via Deepseek's platform API directly, using their V4 flash model, and hooking into a harness like Opencode more than acceptable. Think I've spent maybe $10 over a couple of weeks. I did explore self-hosting models but hardware right now is just too expensive.

Directly at DeepSeek? It was my understanding (but I didn't check) that some other AI operators were providing (some of?) DeepSeek's model for cheaper prices. Still, that's interesting. What do you get for that price? Only coding, or also e.g. image generation?

DeepSeek API gave 6x to 8x better caching rate for inputs over OpenRouter (even chosing DeepSeek as provider). And some of the cheaper providers are using FP4 quantizations.

https://openrouter.ai/deepseek/deepseek-v4-flash-20260423#pr...

After complaints the cached read is not listed anymore in that page, you have to click one by one. All providers for DeepSeek V4 Flash charge ~$0.02 while DeepSeek provider is $0.0028. For coding this is huge as caching often gets in the range of 90 to 99%. But OpenRouter messes your caching so don't use it. And it seems to be a VC-backed closed middle-man company, not open source or open anything.

Re: AI coding at home without going broke

#182

Post by apsurd » Sat, Jun 13, 2026, 9:33 PM UTC

dofm wrote: post #165

Barbing wrote: post #126

Earlier quoted context omitted.

You sound awesome. Just venting? (b/c curious if friends can fill your heart abundantly, & we know we're never too old to make new friends!) > dreading Even avoiding political headlines (OK, at least articles), plenty of cause for dread, so I keep re-focusing to avoid despair. Easier said than done innit! Can't kill my hope for the future though. One day, all the good stuff shall prevail (morality, intelligence, love…

I must say I am not quite just venting. I have been struggling severely with burnout for a couple of years and as I work to fix it by myself ultimately, and get back who I was, the awful thing is finding out that the industry is so utterly and completely different anyway. So in my fight back I decided that I needed to re-centre myself; learn how these tools can help me personally return to productivity, try to get th…

I'm younger, but not by much and I too feel instinctively sad by how abruptly the entire industry has changed. And there's no going back. It's because I'm a craftsmen, I care about the code. And you learn in your career that it's a bad idea to care about the code, especially in a business context, which one's career is very much trapped in the business context.

I care about the code because the code is the product interface to the people working on it, my peers and team. The UX around that code affects us every day, every hour. We should care about it! It took me a decade to realize caring about the code is not bad, it's just a dualism we have to hold: two truths. The code is a means to an end, the outcome and end-user value is the only thing that matters, it's true! Also the code matters. The code is a manifestation of the effort and human attention toward an interface that becomes a product that produces business value for people.

Writing code is changed forever. And I'm saddened by it because I spent so much intimate time and attention writing code. I felt proud and it was beautiful to me, the code itself, the APIs created, and the end user state. (I'm a product developer, and believe it or not, I even enjoy CSS). But also the code is just code. AI writes code. And everyone is rightfully so losing their minds over it all. My hours "coding" are changed forever.

But I fully believe the pendulum will swing back to what has always been true. It's not a failure of AI. It's just what has always been true: creating useful and usable product experiences, for people, is hard. It's a very hard iterative feedback loop with experiential, tacit, actions and actors in real life.

So I think, we're ok. The variance is high and wild, but, it's all good, it's all still ok.

Thanks for your writing, I enjoyed it. (edit: TLDR I think you're product person caught in backend-dev circles. Human-centric, make things for people. In this world, AI is more obviously a tool. On the other side of the pool, the more backend-heavy the dev, the more everything is just one skill file away: marketing, sales, UX, design, writing, strategy, consciousness.)

Re: AI coding at home without going broke

#183

Post by m3kw9 » Sat, Jun 13, 2026, 9:35 PM UTC

That’s easy, just use the plus plan and learn how to prompt efficiently

Re: AI coding at home without going broke

#184

Post by cubano » Sat, Jun 13, 2026, 9:49 PM UTC

sunrunner wrote: post #166

frizlab wrote: post #152

Earlier quoted context omitted.

I recently in $COMPANY had a coworker try fable to do a refactor where not breaking anything was the game. It broke something at the first PR. I think we’re not there yet.

I've found that adding "Make no mistakes." to my prompt usually helps with this kind of problem...

perhaps simply threatening to fire it would also do the trick...it sure has worked well on us for a long time now.

Re: AI coding at home without going broke

#185

Post by bachmeier » Sat, Jun 13, 2026, 9:51 PM UTC

sbochins wrote: post #179

bachmeier wrote: post #89

> The upfront cost is steep and the models you can actually run at home are weaker than what the frontier labs ship, so this only pays off if you can keep the rig busy with long running tasks where a slower, cheaper model grinds away overnight. Most people can’t keep a home machine that loaded, and the hardware you buy today may look like a bad bet in a year. Oh, so this is not a post about AI coding at home. It's ab…

There are certain things you can leave running for a while. I think the distinction between vibe coding and hitl based coding routines will blur as workflows prove themselves and models become smarter and less expensive. Most of the best engineers I know have transitioned a lot more into vibe coding this year. The possibilities are much better nowadays.

> I think the distinction between vibe coding and hitl based coding routines will blur as workflows prove themselves

There's far less need for what the author refers to as frontier models as soon as you move away from vibe coding to filling in the gaps that you don't want to write yourself. The author doesn't even consider Gemini models to be frontier.

> models become smarter and less expensive

That's optimistic. They might become smarter but I don't see any market forces in the next few years that will make them cheaper.

Re: AI coding at home without going broke

#186

Post by cubano » Sat, Jun 13, 2026, 9:54 PM UTC

dpcan wrote: post #136

I cannot figure out what people are doing to spend all this money. I have used a $60 per month Cursor plan on auto, and have never come close to using up my included usage, and I probably have it planning and coding and working for me all through the evenings 4 nights a week. What on earth are people doing differently that it's costing them so much? Maybe enabling on-demand usage or other paid models, or on higher mo…

> and they tell us they use $2K per month in tokens with their current employers...

perhaps they are simply trying to impress you with their mad prompting skills and like, what self-respecting engineer would be caught dead using less then $2k/month?

giving the context of your interaction with those people, it probably is the simplest answer to your rather baffling question. for the life of me the idea of using $2k/month doesn't even seem possible unless your telling it to waste credits.

Re: AI coding at home without going broke

#187

Post by Random09 » Sat, Jun 13, 2026, 10:02 PM UTC

gabriel-uribe wrote: post #174

Random09 wrote: post #159

Earlier quoted context omitted.

What I mean is a script that can look through the logs. They are known and deterministic (if you properly handle errors) and you can analyze them statistically. If you don't know what logs your app is outputting, then you have a bigger problem in your hands tbh.

Deterministic scripts are awesome, and they certainly power my internal dashboards. But I'm a human - I will miss things. I maintain too many apps to have entire codebases in memory at this point. Or to continue monitoring all these streams. Logging is cheap - I log as much as possible because an AI will scan it for me. I just want scoped pull requests to review proactively against the slew of things that can happen…

It seems you could use a human instead then? If you have so many apps, you could hire a junior to help you. There is additional satisfaction of bringing new person to the IT too.

Re: AI coding at home without going broke

#188

Post by lemming » Sat, Jun 13, 2026, 10:02 PM UTC

sunrunner wrote: post #166

frizlab wrote: post #152

Earlier quoted context omitted.

I recently in $COMPANY had a coworker try fable to do a refactor where not breaking anything was the game. It broke something at the first PR. I think we’re not there yet.

I've found that adding "Make no mistakes." to my prompt usually helps with this kind of problem...

Or if the code is really important, sometimes even “please make no mistakes” is necessary.

Re: AI coding at home without going broke

#189

Post by nunez » Sat, Jun 13, 2026, 10:03 PM UTC

> The second is to skip the hardware and rent those same open source models from a provider at API rates. For most people this is the right call. You avoid putting thousands of dollars on one GPU setup while configurations are still in flux, you skip the work of squeezing long running performance out of an open model, and you can switch to whatever is cheaper or better next month without reselling a box. Something like OpenRouter makes the move close to a one line change.

This will probably become the only option as the companies that publish open weights stop doing that. Very very few people have enough hardware to train/fine-tune at home.

Re: AI coding at home without going broke

#190

Post by dudisubekti » Sat, Jun 13, 2026, 10:04 PM UTC

CamperBob2 wrote: post #71

theossuary wrote: post #49

Earlier quoted context omitted.

I truly think by 2028 we'll have integrated chip systems that'll be able to run opus 4.8 level models at ~500 watts at acceptable performance. Honestly I think now is the worst time to invest in AI hardware. Get your harness ready and processes perfected with hosted models, and wait a few years to buy hardware to transition to running models locally

Honestly I think now is the worst time to invest in AI hardware. That position is not without its own risks, though. Maybe Opus 4.8 will run on a single chip by 2028... and maybe you won't be allowed to touch it. And what if Xi makes a play for Taiwan? That would be stupid, but so was invading Ukraine with tanks from Temu, and it still happened.

Other than Taiwan declaring independence, I don't see any reason why China will rush to take the island.

At the very least they would wait until they cracked EUV and mass-produce the chips, and that is still 4-5 years away at the earliest.