Honestly, my accessibility on my apps/websites is much better now with AI because you can just tell AI to do it (and run automated tests to validate it worked) vs not doing it at all for a small side project with 2 users.
Anthropic is really trying to burn all that goodwill they worked up by raising prices, reducing limits and making it impossible to know what the actual policies are.
> a solid product and company can withstand online controversy
A product with a massive moat. Switching from Claude to another competitor is insanely easy and without much loss of quality. Until they’ve built their moat, burning goodwill is foolish.
What’s different is it’s probably required due to the cash that’s being burnt to operate. They can’t afford to keep offering so much for so little revenue.
If you want LLMs to continue to be offered we have to get to a point where the providers are taking in more money than they are spending hosting them. And we still aren't there (or even close).
Nope. They're losing money on straight inference (you may be thinking of the interview where Dario described a hypothetical company that was positive margin). The only way they can make it look like they're making money on inference is by calling the ongoing reinforcement training of the currently-served model a capital rather than operational expense, which is both absurd and will absolutely not work for an IPO.
Inference, in and of itself, can't be completely unprofitable. Unless you're purely talking about Anthropic?
But
> If you want LLMs to continue to be offered we have to get to a point where the providers are taking in more money than they are spending hosting them
Suggests you just mean in general, as a category, every provider is taking a loss. That seems implausible. Every provider on OpenRouter is giving away inference at a loss? For what purpose?
For the same reason that Amazon operated at a loss for two decades and Uber operated at a loss for a decade and a half. The problem is the free money hose isn't running anymore.
The open models may not be as great but maybe these are good enough. AI users can switch when the prices rise before it becomes sustainable for (some) of the large LLM providers.
Currently it costs so much more to host an open model than it costs to subscribe to a much better hosted model. Which suggests it’s being massively subsidised still.
For a lot of tasks smaller models work fine, though. Nowadays the problem is less model quality/speed, but more that it's a bit annoying to mix it in one workflow, with easy switching.
I'm currently making an effort to switch to local for stuff that can be local - initially stand alone tasks, longer term a nice harness for mixing. One example would be OCR/image description - I have hooks from dired to throw an image to local translategemma 27b which extracts the text, translates it to english, as necessary, adds a picture description, and - if it feels like - extra context. Works perfectly fine on my macbook.
Another example would be generating documentation - local qwen3 coder with a 256k context window does a great job at going through a codebase to check what is and isn't documented, and prepare a draft. I still replace pretty much all of the text - but it's good at collecting the technical details.
I haven’t tried it yet, but Rapid MLX has a neat feature for automatic model switching. It runs a local model using Apple’s MLX framework, then “falls forward” to the cloud dynamically based on usage patterns:
> Smart Cloud Routing
>
> Large-context requests auto-route to a cloud LLM (GPT-5, Claude, etc.) when local prefill would be slow. Routing based on new tokens after cache hit. --cloud-model openai/gpt-5 --cloud-threshold 20000
I've found MiniMax 2.7 pretty decent and even pay-as-you-go on OpenRouter, it's $0.30/mt in, and $1.20/mt out you can get some pretty heavy usage for between $5-$10. Their token subscription is heavily subsidized, but even if it goes up or away, its pretty decent. I'm pretty hopeful for these openweight models to become affordable at good enough performance.
Rapid MLX team has done some interesting benchmarking that suggests Qwopus 27B is pretty solid. Their tool includes benchmarking features so you can evaluate your own setup.
Edit: I’d also consider waiting for WWDC, they are supposed to be launching the new Mac Studio, an even if you don’t get it, you might be able to snag older models for cheaper
I see the current situation as a plus. I get SOTA models for dumping prices. And once the public providers go up with their pricing, I will be able to switch to local AI because open models have improved so much.
Like with all new products. It takes time to let the market do its work. See if from a positive side. The demand for more and faster and bigger hardware is finally back after 15 years of dormancy. Finally we can see 128gb default memory or 64gb videocards in 2 years from now.
A guy from Meta interviewing at BBC a few years ago claimed that every school child in India was going to have the metaverse VR or they'd be left behind in their education, so every family was certainly going to pony up the money.
Somethings not adding up. Why is Amazon making financial plans for the next decade based on continued OpenAI spending but you’re saying AI providers like OpenAI and Anthropic aren’t even close to being profitable, so how can they last a decade or more?
That's the interesting question, right? Because if this unwinds during a period of external inflation (say, because of a big war and energy shortage) then even the Bernanke would say helicopter money won't work
They probably aren’t planning on making the money on consumer subscriptions. Any price is viable as long as the user can get more value out of it than they spend.
What shareholders, Anthropic is a money burning pit. Not to the same extent as OpenAI, but both will struggle hard to actually turn a profit some day, let alone make back the massive investments they've received.
Not that they don't bring value, I'm just not convinced they'll be able to sell their products in a sticky enough way to make up the prices they'll have to extract to make up for the absurd costs.
>> both will struggle hard to actually turn a profit some day, let alone make back the massive investments they've received.
I'd agree with you, except I've heard this argument before. Amazon, Google, Facebook all burned lots of cash, and folks were convinced they would fail.
On the other hand plenty burned cash and did fail. So could go either way.
I expect, once the market consolidates to 2 big engines, they'll make bonkers money. There will be winners and losers. But I can't tell you which is which yet.
I’m not sure there will be consolidation. There’s too much room for specialization and even when the models are trained to do the same task they have very different qualities and their own strengths and weaknesses. You can’t just swap one for the other. If anything, as hardware improves I’d expect even more models and providers to become available. There’s already an ocean of fine tuned and merged models.
$20B ARR or so reported added in Q1 doesn’t sound particularly bad, they’ll raise effective prices some more while Claude diffuses into the economy, sounds like a money printer. The issue is they’re compute constrained on the supply side to grow faster…
> $20B ARR or so reported added in Q1 doesn’t sound particularly bad
Unless you compare with the reported cash burn or projected losses.
> they’ll raise effective prices some more while Claude diffuses into the economy, sounds like a money printer
But the problem is, they have no moat. Even if Claude diffuses into the economy (still to be seen how much it can effectively penetrate sectors other than engineering, spam, marketing/communications), there is no moat, all providers are interchangeable. If Antrhopic raise the prices too much, switch out to the OpenAI equivalent products.
I disagree very strongly with this, both anecdotally and in the data - subscriptions are growing in all frontier providers; anecdata is right here in HN when you look around almost everyone is talking about CC, codex is a distant second, and completely anecdotally I personally strictly prefer GPT 5.3+ models for backend work and Opus for frontend; Gemini reviews everything that touches concurrency or SQL and finds issues the other models miss.
My general opinion is that models cannot be replaceable, because a model which can replace every other provider must excel at everything all specialist models excel at and that is impossible to serve at scale economically. IOW everyone will have at least two subscriptions to different frontier labs and more likely three.
You're actually reinforcing my point. Models are interchangable and easy to switch between to adjust based on needs and costs. That means that no individual model / model provider has any sort of serious moat.
If tomorrow Kimi release a model better at something, you'd switch to it.
It's likely that Chinese models will get regulatory knee-capped at some point, and the domestic labs all have pretty common costs they need to make up. This creates an environment where they match each other as prices climb. Unless Google/Meta suffocates the startups since they have actual cash flow that is non-AI.
Sure you can go local, but lets be real, that would be <1% of users.
I postulate in practice this won't matter since the space of use cases is so large if Kimi released the absolutely best model at everything they wouldn't be able to serve it (c.f. Mythos).
Aren't they just doing what Hacker News was trying to tell them to do? That AI is useful but not sure if sustainable. Now they're increasing prices and decreasing tokens and you guys are pissed off.
Failed to build one but my advice would be to focus very narrow. In your case start with literally between two cities. Also, fake supply by either paying people to do the trip (in addition to the normal payment on the platform) or literally do it yourself once per week. Focus on supply. It will be way harder to get.
Haha no, I been in 90c saunas many times. Can I stay there for a long time? Heck no. But some people can and it doesn’t kill you (maybe if you have some preexisting condition)
reply