Comment by drevil-v2
6 hours ago
The damage is done. You cannot build a business critical function on top of American SOTA frontier model. Especially not with the current crew in charge.
Now whether AI tech is in the same league as say Nuclear tech and therefore by any reasonable standard should be regulated is a different question.
We hit the slippery slope on a random day in June 2026 and there is no putting the genie back in the bottle. Any exec or manager that puts load bearing weight on top of Anthropic/OpenAI/Google/AmericanCorp frontier model deserves the stress.
> The damage is done. You cannot build a business critical function on top of American SOTA frontier model. Especially not with the current crew in charge.
The switching costs of changing LLM providers is as low as it gets. All the individuals and startups I know try different models all of the time, even down to the level of choosing which provider to use based on the task. Bigger companies move slower but only because they have lawyers and teams negotiating contracts, not because there is a technical reason that it's hard to switch.
Companies have dealt with supply chain unpredictability by having multiple providers and switching between them since forever. It's infinitely easier to switch LLM providers than it is to deal with physical supply chain uncertainty.
Exactly!
Even if you won't be able to use some model tomorrow, you can still make money by using it today!
And in the age of limited compute, spiky workloads and constant outages, building a mechanism to fallback to a weaker model when your primary choice isn't available is smart anyway.
I'm a small software business owner in Europe. I have to assume my competition is willing to pay for any business advantage they can get. And so I also have to pay for the SOTA model, whatever it is.
No you don't; it's often overkill to use the SOTA models. People want SOTA because it's shiny, but there are a lot of tasks where it's cheaper and more efficient to use other models.
The good news (for you and most everyone other than the current leading AI companies), the gap between the SOTA and the near-frontiers is getting smaller every week or two. The leading Chinese models are only a few months behind now (GLM 5.2 tickles the tail of GPT 5.3 or 5.4 and Opus 4.6, according to benchmarks and the vibes among heavy users who've spent some time with it), where they were a couple of years behind a year ago.
4.6 was released at the beginning of February, so if the Chinese models only "tickle its tail," that means they're >5 months behind.
6 replies →
This is nonsense.
The gap between Chinese models and American frontier models is estimated at 10 months by Anthropic themselves, and it's growing.
China has no flywheel for long-form agentic traces like Claude Code and its telemetry over its userbase (no one uses the Chinese harnesses yet). Most Chinese models are forced to price themselves significantly below cost to compete with the huge demand for bootleg claude tokens, because they're that much worse.
3 replies →
Another day, more cope on this subject from many posters on here...
I don't know if you write software for your own products or if you code for your customers. Anyway, are you going to compete on the speed of your code writing AI or on deploying the features your customers need? One useful feature is better than a hundred ones nobody really care about. And a good relationship with customers is better than any feature.
Example. Yesterday I listened the technical lead of a customer of mine digging himself into a hole by not understanding what it would mean exposing AWS EFS to their on premise server over NFS. It was just too many unknown unknowns for him and he had no time to ask the AI (and even if he did I'm not sure that he could understand.) His boss, which actually used NFS, had to stop him. I didn't speak a word.
So, he could have coded the migration of a server from AWS to on premise, asked Claude to write also all the configuration scripts and policies but then what?
This is a great recipe for going out of business.
If the competitive risk is real, then are choosing between supplier risk (AI model access) and competitive risk.
When there isn't a zero-risk option, the question becomes which risk is smaller.
2 replies →
Any competitive business will accept this risk if it gives them any type of edge no matter the duration of that edge. This is no different that using an exotic raw material.
3 replies →
And thus, capitalism continues to roll on. Businesses are suppose to go out of business, its a feature.
4 replies →
What concrete business advantage are you getting from LLMs?
Speed.
48 replies →
If you can't figure out what model to use your business is already dead.
This thinking that every task must be stuffed into the most 'advanced' (expensive) model out there is idiotic, and it's not only you unfortunately.
At $JOB I have warned higher ups we should try to keep our expenditure under control, educate people that document slinging doesn't require Fable every time and demo the capabilities of the cheaper models, and been snubbed for it. When Fable is available once again our bill is going to be eye watering, relative to what it should be.
This! I've found that for most coding, Sonnet is pretty good as it is. Yeah, you might need to finesse your prompt a bit more, and you'll probably be spending a bit more time on the computer, rather than a more hands-off approach, but at the end of the day, you'll save a lot more simply because you're using a good-enough model.
If you're the one-shotting type, obviously then Fable might be useful, but I think only marginally. You don't need to bring a MANPADS to a duel at high noon.
1 reply →
do you think your current operation and niche is so optimized that not using Fable would put you out of business? Or is this a hope that using Fable will allow you to stay in business?
I am on track to commoditize my niche industry, and I hope I can do it before anyone else beats me to it. I'm working at panic speeds.
1 reply →
So the panic generators ("You will be left behind!") are winning. Creating a sense of urgency that makes you switch off the higher rational functions is a key element in every successful scam.
Unless you have concrete evidence via evals that SOTA is actually needed, you’re just buying into the hype.
Nonsense. Do you buy state of the art pens, pencils, printers, paper, computers, disks, etc.? No. You buy whatever is the best value for the case at hand. That’s often not the SOTA option.
Sure but that's orthogonal.
Yes you use the right tool for the job.
But if the job requires the best intelligence you can get with an LLM, then you use that.
Taking as an assumption that the quality of your product is a function of the quality of the inference you are using: if you use an inferior model because "what if it gets export controlled again" and your competitors don't, then your competitors are likely to win.
If you don't need frontier models for you job then this is all moot, but the thread started with
> You cannot build a business critical function on top of American SOTA frontier model
Which is silly. HN likes to roleplay bringing everythgin "business critical" in house because sometimes vendors mess up. Self host, don't use the cloud, run open models locally, built redundant supply chains in case of another covid, etc etc. Sometimes the risk is real, but most of the time the risk is rare and the cost of an interruption event is less than the cost of bringing everything in house or using lower quality vendors "just in case"
Wouldn’t you just have fallbacks? Today’s frontier models are just better than the other models, they don’t really have a ton of entirely unique abilities that can’t be replicated with more time and effort.
So you use the frontier model, then when you can’t you accept things are less efficient. The alternative (right now) is to be less efficient all the time, I don’t see any advantage to that.
Yeah. It’s not the end of the world.
But, it is a big own goal, because once you invest in building evals for your internal use-case, 1) it’s easier to switch your model to whatever is cheapest, and 2) it’s way easier to fine-tune an oss model.
Evals are annoying to build and most companies were fine to rest on vibes. Now many companies have to do the work for insurance.
The real problem is the White House just making up the rules as it goes. No laws. No predictably for the markets.
A week or so pause from seemingly legitimate cyber security concerns isn’t cause for panic. But it should be backed by laws that describe what that process should be. That would put the market at ease
There’s no optimal answer.
The reality is this is world-ending technology and absolutely nobody knows what to do or can even agree that the problem exists.
> "The reality is this is world-ending technology and absolutely nobody knows what to do or can even agree that the problem exists."
The reality is that the "people in power" believe it is "world-ending technology" and will therefore use it in world-ending ways. People are absolutely 100% the danger here, not the technology.
Photosynthesis once ended the world
> You cannot build a business critical function on top of American SOTA frontier model.
Yes 1000%, please, all my European competition please don't use mythos whatever you do it's total USA trash and the Chinese models work better anyway.
[flagged]
Read the guidelines, you can make your point without calling people "suckers".
> When disagreeing, please reply to the argument instead of calling names. "That is idiotic; 1 + 1 is 2, not 3" can be shortened to "1 + 1 is 2, not 3."
https://news.ycombinator.com/newsguidelines.html
I predict we all be using the hell out of fable until the next great model comes around and in two weeks we won’t be talking about the export controls anymore. We just don’t have the attention span.
Nobody should be putting loadbearing weight on Amazon or Microsoft with their ruthless monopoly ambitions, yet here we are
> I predict we all be using the hell out of fable until the next great model comes around and in two weeks we won’t be talking about the export controls anymore.
Until it goes down, or Anthropic raises prices again.
Fable is already expensive to use compared to GLM and they want you to use the API as much as possible so you get a worse deal.
Why would you compare Fable to GLM? What a bizarre comparison. They're at least two generations/tiers of intelligence apart. GLM is great and I use it often but it might as well be Sonnet when compared to Fable.
Feels like a leap. This kind of move was always possible. It's possible China stops publishing their frontier too. US could lock down access to Nvidia latest hw of scale even if you intended to do open models. Then what? Say amd or bust? The best you could do going solo (i.e no nation-state interference) is tiny stuff that you can run on commercial stuff. But that is seriously limited / slow in comparison. You either have to do dumb and fast, or smart and slow IME for these self-hosted things that aren't on the beefy racks.
The real answer is you should never build your business on ANY specific model. As usual avoid lock-in and switch when you need to.
Nah, people will still pay, as many if not most consumers truly do have a short memory. And like other comments say, imagine everyone is using Fable and you are not, you will quickly fall behind, per the Red Queen hypothesis.
LLMs are still easily replaceable. If the SOTA frontier model provides meaningful impact for your critical business function, then worse case you flip the switch to the next most capable model.
Unequivocally false. Models have different behaviors, parameters, tool calling templates, etc. The providers publish extensive documentation on all of this. Yes, you can take the quick way and swap a model, but it will not run at its full potential until you adapt your workflows to it.
Or you can use universal harness that is not tied to a specific model (there are many available now, such as OpenCode, Pi..).
I think this is black and white thinking. Fable and US AI is not unique technology. It’s just marginally better than open source tech at 10 times the price. You can swap out the models at will, they are pretty much fungible. If your use case can pay for a best in class model then you will pay for it no matter the bogeymen. If your best in class model becomes unavailable, you switch to the next best model for a very minor performance degradation. I really doubt this will deter anyone from using American AI.
This won't age well. You just need to code in a way that has fallbacks. Whether that is to older models, different companies. It's going to be a commodity (if it isn't already).
Most companies do not model themselves as "building on [AI model du jour]" yet. They model themselves as building products with those tools, which they consider as relatively substitutable.
The EU has totally and utterly failed when it comes to frontier AI. They are out of the running, they won't catch up in time for superintelligence. There literally is not enough compute for sale in the world for them to do so.
They crippled their own domestic entities with the AI Act. (see the Mistral CEO's rant.)
If you want to use frontier models until then, you're gonna use what's available, and that's US models.
> You cannot build a business critical function on top of American SOTA frontier model.
Yes I can!
Better to fix it now than tomorrow.
The Damage: Now every time Claude does something stupid or trashes your code, developers in the back of their mind will think, is Claude sabotaging me on purpose? [1] Trust is hard to gain. Easy to lose. And harder to get back. Models will converge. Trust won't.
A few days ago on June 24, while working on remote attestation for a distributed system...
What are we to think? Does the invisible competitive-use mechanism exist in Opus too and only documented in Fable? How long has it existed? Is it still in effect? -- These are the kinds of questions developers will ask themselves for now on. This is why it was one of the stupidest things Anthropic could have done. Developers will now question everything and rightly so. There's no attestation protocol for that. How will they know?
[1] "In light of the ability of recent models to accelerate their own development, we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms.
Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts,these safeguards will not be visible to the user. Fable 5 will not fall back to a differentmodel. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT). These interventions will not affect the vast majority of coding work. We estimate they will impact ~0.03% of traffic, concentrated in fewer than 0.1% of organizations. When these interventions are active, we expect them to have minimal behavioral impact on the model except to limit its effectiveness in developing frontier LLMs. Claude will still respond helpfully to user requests. We’ll continue to improve the precision of our detection methods following the launch of this model."
Source: https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3...
They undid this after the backlash
[dead]
[flagged]
that is exactly why downvote exist.
That is part of why downvotes exist. They also exist for personal attacks, off topic tangents, posts that don't make sense, trolling, advertisements, AI generated content, and other such things we don't want to see here.
But "downvote for disagreement" is a legitimate use. I personally tried to tell someone that it wasn't, and I got corrected by dang.
1 reply →
composer 2.5 is all you really need don’t be so dramatic
Nobody cares about this temporary "ban" by the US government. If anything it only increased the mystique of the two models.
I think Europe and Canada are just happy not to be frozen out of AI access completely at this point.
All the discussion this week have been about GLM, Qwen, etc. Both over 1000 comments in the last couple days.
https://news.ycombinator.com/item?id=48721903
Of course Anthropic is still relevant, but people have realized they’re not special, and between this and the ID verification thing, they’ve given up a ton of their relevance vs a month ago.
I used Fable 5 for maybe 10 hours in the window when it was available. It was much better than Opus 4.8. And I have found the Opus models to be excellent, but Fable 5 was cranking out incredible research on some data sources I wanted to plumb into my project.
I wouldn't personally pay API pricing for it for my personal projects, but I bet it's going to be absolutely slammed with usage for the next month+.
1 reply →
As everyone knows, Kool-Aid is also just mostly water.
I work in AI / infrastructure and I have never seen as much interest towards investing into sovereignty by actual deciders. Thankfully, at this point I can't see any flip-flopping / change of messaging stopping that train.
In CA/EU over the last ~15 years, one used to be perceived as a bit of a "weird systems person" by just proposing alternatives to the big hyperscalers.
So the Trump administration, hands-down, has been the greatest ally here.
In tandem, I was hoping Anthropic would be keeping "dangerously capable" models banned from "evil Chinese distillers" for as long as possible.
[flagged]
I bought a GLM 1 year subscription and changed my environment variables to use Claude Code... yep the same one that is using stegonography to send details about users to the model. China knows where I live, I'm not getting ripped off or rug pulled on their models either.
How has your experience been so far? Did you previously use Opus? Im curious how the overall "feel" of it is.
No nation is going to willingly release a model that can be used against it. Not even China. The moment they have a Mythos class model, they will go through the same process. The AmericanCorp models are far ahead of any other models so we see this process unfolding through that lens.
No Mythos class model will be allowed to be legally hosted for download on any service. All powerful nations will ban this since safeguards are not guaranteed by shady service providers running these models.
For the Chinese first party providers, they will be forced to implement the same process and safeguards, and they will not be allowed to release the model weights to the public.
Why? Because no sane nation is going to put that kind of capability in the hands of the public only for the public to use that power against that nations best interests.
Save this comment. It is prophesy.
Not great news for nations that want to secure their software.