I think you're overlooking the fact that for long-horizon tasks, even small errors compound over time and can lead to catastrophic outcomes.
For simple queries, we have reached the threshold since the beginning of the year, and models are good enough from every provider to make a meaningful difference between one another. (ChatGPT, Claude, Gemini, Grok, MuseSpark, Kimi, DeepSeek, GLM...)
The real unlock will be, and you can already see it with GPT-5.6 and Fable-5, to delegate complex enough tasks that will take more than 24 hours to get done and they will not lose track. I'm not talking about a loop, but the actual intelligence to recover from these compounding errors that accumulate in dumber models.
We're still a long way from the intelligence needed to let one of these agents go ahead and supervise multiple layers of sub-agents underneath to do complex orchestration. The future looks very promising and exciting. Imagine having the possibility of a Frontier model orchestrating as many sub-agents as needed that are running on cheaper models like DeepSeek.
That doesn’t make sense. It’s not like SOTA models are error free, yet we still use them.
You use Fable 5 right? If that’s good enough for you now, why wouldn’t a Chinese model that’s as good as Fable 5 but at 10% the cost be good enough in 6 months?
I think we put up with Fable's occasional hiccups because there's nothing better at the moment.
I use Claude Code semi-heavily for my small business, and the $100/mo I pay for that is a rounding error compared to the value it provides.
If I can avoid spending an hour or two "massaging" the output from a lower-end model once, or it avoids introducing one load-bearing (sorry, couldn't resist) bug, then that's the entire $100 right there.
Hell, you could argue that the best "coding model" that we have at the moment is the human brain, and people will gladly pay $10,000/mo for one of them.
Arguing over $20 vs $100 for something that actually puts in work just seems insane to me.
> I think we put up with Fable's occasional hiccups because there's nothing better at the moment.
Which was an argument for using every less powerful model since the moment they got useful, right?
When was that? Opus 4.5 maybe? Let's say Opus 4.5 for the sake of the argument. So back then we were like "DeepSeek is not good enough, I need Opus 4.5". Now DeepSeek is better than Opus 4.5. So if Opus 4.5 was good enough back then, DeepSeek is better than that now.
Sure, it's always nicer to have a slightly better model. But the price difference starts mattering a lot more when all the models are already sufficiently good.
To put actual numbers on it, since using AI to start solving all kinds of bottlenecks/inefficiencies in our small business, we've seen monthly net profit go up by around $4,000 USD. These are semi-permanent fixes, and the tech is only partially deployed. I am the only one using it, and I only use it part time.
We've just spun up our first Hermes agent, with direct API access to our main inventory system and that's expected to find another few grand per month in misallocation/inefficiency.
I wouldn't be surprised if we were doing more like $10k/mo higher in 6-9 months' time.
When you're talking about numbers like this, the fact that one AI is $100/mo and another is $10/mo or $40/mo doesn't matter. They could make GLM-5.2, or any other Opus 4.5-class model free and it still wouldn't make sense to deploy in a commercial context.
The other angle I'd approach things from is that Opus 4.5 (and I'd agree with you that that model was the saddle point) was "good enough" for the types of things we were asking it to do back then, but as the models have become more capable the tasks we're asking them to do have also expanded with it.
I know I've personally gone from "hey can fix this race condition with a Redis mutex" 6 months ago to "Independently redesign this full embedded USB stack and QA it end-to-end, working around a specific Kernel bug in macOS Tahoe that requires decompilation to find the source of, while keeping in mind the constraints of our 8-bit AVR chip from 2011" now.
But that said, yes, maybe in 5 years' time we will reach an "intelligence saturation" where the average person won't be able to even conceive of how to use the new SOTA.
I think we're even starting to reach that saturation point now for a lot of people. In my industry (law) plenty of people have tried CoPilot once or twice, or tried ChatGPT a year ago, and as a result have basically dismissed AI as being useless. The setup required to be able to get it to do end to end tasks to your liking is also substantially more work than most people are willing to put in.
I think HN doesn't really understand fixed costs, I spend a few hundred dollars per day on Fable and the costs are irrelevant compared to what we make.
We are obviously limiting access to be read-only for anything customer-facing (it will be able to put recommendations in the dashboard but not actually change things directly) but I think the point of GP's comment was literally just to get a reaction.
No, the point is that with human in the loop the downside is (usually!) rather limited, as common sense would stop obvious fuckups (ok, not always, but still).
With an agent (especially incompetently employed), the danger of unwittingly destroying your company (or at least, the crucial data/reputation) is rather higher. We are notoriously bad at estimating the downside risks in complex systems.
The most obvious case is the downside risks in complex financial constructs... things look great for a while ... until a sudden surprising collapse arrives and totally destroys all the upside you think you have created.
There is a tradeoff to be considered between the utility gained from using the stuff, minus the risk severity/likelihood, plus available mitigations. As someone else posted, none of us are in a position to make that balanced post because we don't know if the guy has an airgapped backup or not etc.
In any case, GP's post was not such a balanced consideration; it was just parroting a beloved risk-aversion meme that can easily be deployed against building anything (what if the building falls on top of someone?) or even leaving home to go to work ("travelling in a hunk of steel at lethal speeds – let me assure you that absolutely nothing can go wrong here, mate.")
What I find tiresome about that meme is the presumption that "something can go wrong" is useful input on its own. It's not. Mistakes are made all the time, the only way to avoid that is to stop breathing. Even in the process of me standing up and going to the loo, something can go wrong.
If the guy wants to make a case that it's too dangerous for the expected benefits, he has to actually make that case. Saying "risk exists" with no elaboration is a waste of HTML. "something can go wrong" every time he swallows food, yet mysteriously he still does it.
(the suicide analogies may seem mean-spirited, but I kind of mean it. If you consider every action primarily from a standpoint of "what harm or irreversible change can result from this", the only permissible path is to do nothing. To be moral is to be as close as possible to a rock or another inanimate object.)
The question low cost models will create: Why would you massage output?
Fable 5 is still going to mess things up at any sufficient complexity. The advantage of low cost models with "good enough" intelligence is they can recursively correct. Why? Because it is cheap. Proper requirements and tests and subagents take away increasing amounts of work, at a cost that is not prohibitive.
If you are reviewing code manually you might consider Fable 5 a worse option. As it articulates itself with higher confidence and you already know it is capable, you are may be more likely to miss a mistake. You know to be on guard with a junior engineer. Reviewing a senior who suddenly makes some weird stochastic mistake can be a lot harder. It would be like if the smartest human engineer you knew was capable of some random brainfart in the middle of their massive diff. Imo, much harder to deal with.
Of course, we should keep in mind Fable 5 is only expensive today. It will be cheaper in the future. Autonomous, recursive prompting and improvement is the clear end state. Especially for entities that will always have the budget for that at the SOTA frontier.
It's far from significant, it's partially doubled during peak hours. They could 16x it and it would still be two orders of magnitude better value than OAI's $200/mo plan.
It's that good. They are far from capacity limited, and even if they were, you can rent a single MI300X from somewhere like Hot Aisle and get more tk/s than you'll be able to use.
How does it compare to 5.6 Luna after the permanent 80% price cut?
That one is dirt cheap at API pricing, I can't imagine quota is going to be a concern on the $200 subscription, which in my opinion easily supports full time use of 5.6 Sol on xhigh.
It's still laughable. They blew it. I'm not sure they could even pay me to use their models at this point (and I don't mean via employment, meta and its refugees are permabanned to me). My experience over the past few months with open models has me seriously entertaining moving east.
A couple of very talented friends were uttering curses upon the entire bloodline of whoever convinced them to try letting sol xhigh do serious work. Deepseek cleaned it up for a fraction of $20.
> Didn’t deepseek recently announce prices will go up significantly?
They also previously said prices will go down significantly once they get a hold of the upcoming Huawei chips (later this year).
Prices are going up just because they can. It can easily come back down. They aren't strained by some IPO / VCs requiring them to 1000x their earnings.
I don’t think the industry knows how to price this stuff. Deepseek is great (I’m running it on a RTX 6000 pro setup) but it’s nothing like Fable. It’s still strongly human-in-the-loop which is fine, until you experience how good these models can be.
Think about it this way.
Let’s say you could buy an LLM that gets things right 98% of the time. But there’s another LLM that’s 100x the price but gets things right 99.9% of the time. To the lay person this sounds trivial but to a serious business this intelligence gap could represent millions, or billions of dollars.
If that’s the case businesses would be seeing millions to billions of profit gain (or cost reduction) in the past 4 months as they went from Opus 4.6 to Fable 5.
But that’s simply not the case. It’s very clear that vast majority of the business do not generate additional value from incremental intelligence gain from these models.
There is a reason why Chinese open weight models are now popular even in American enterprises, because CTOs realize that they are indeed good enough.
I work for a FAANG, and have my own personal projects for which I use the Chinese models. and the big models do indeed save/make us a lot of money.
The Chinese models are not good enough for anything other than pair programming, which is just a very last-gen way of using agents.
And when the big US models get better we will move with them. Until we stop seeing returns there is no "good enough", I don't know why this is so hard for HN to understand.
Fable has only been out for a month but somehow everyone is supposed to have moved to a completely different way of working that supposedly only works for Fable and nothing else…
This stuff is just obvious to anyone working with the latest models. Fully autonomous agents are a game changer. Having to pair program with one is indeed "last gen", I haven't done that for a month and I won't ever be doing that again in my life, outside of personal projects.
> This stuff is just obvious to anyone working with the latest models. Fully autonomous agents are a game changer.
This kind of takes makes me cringe. Why don't you go back to LinkedIn?
I have no idea what you mean by “pair program with an agent”, but Opus have been able of autonomous coding since last November, and with any half-decent harness even local Qwen3.5 was able to do so 6 months ago.
Fable is a stronger model, which means it can solve harder tasks but it's also over-hyped, because only a small fraction of task is hard enough to be Fable-worthy.
I use Kimi K3, Opus 5, and Fable on a daily basis.
Fable is the only one that reliably one shots complex changes and makes the right design choices. Everything else requires handholding.
I can let Fable loose on a 12+ hour (for AI) task and it will have performed it flawlessly when I come back the next day. K3 and Opus are not like this.
And no, our harness is not the limiting factor here.
Unless you are very cash strapped, Fable is a very nominal fixed cost compared to the benefit of what it offers (fully autonomous agents, and no longer needing to pair program with one).
And even it isn't "enough". I can very clearly see myself using more advanced agents to move up the abstraction ladder.
For businesses that have actual problems to solve, I see them investing in the frontier for a good bit longer, probably until we have AGI that can replace employees, maybe even a bit after.
This is why I find the "good enough" arguments silly. Like, the usefulness of an AI tops out to you when you can pair program with it? Seriously? You cannot envision ways in which more advanced AI enables you to do more, better? That's bizarre to me. I don't ever see myself running out of problems to solve.
I use Chinese models, even smaller local ones, for much more than pair programming. If we are talking about deepseek v4 flash, which is basically a frontier model, it is much more capable than the local models I run on my MacBook Pro. The only issue really is finding the right harness.
I do have a way of correcting through redundancy, though. If you are just vibe coding, you need to use the most capable model you can find and even then it might not be good enough.
> If you are just vibe coding, you need to use the most capable model you can find and even then it might not be good enough
I mean, this proves my point. Better models enable you to get more done. With Fable, 80% of the time, I no longer have chat with an agent over the details of a PR. I give it an outcome and it gets done. This means I can work on much more with the limited time I have.
And I don't see this ending. When better models come out that take that from 80% to 99.x%, I will have that better model manage teams of other models and move up the abstraction layer.
If models get even better than that, perhaps I stop reviewing PRs entirely. Maybe normies can start using agents to build real things.
Unless your business doesn't have many problems to solve and isn't in a competitive environment, it will benefit from using the best models.
If you don’t have a way to automatically check the results via redundancy, you need a really good model, since even 99.9% reliability is going to cause slot of headaches. If you do have a way to automatically check results, then you can use something that fits in your computer and was produced 3 years ago.
My point is you don’t need the best model if you just put in QA processes that can be done by models also. And if you don’t have that, the model is probably not going to be good enough.
> The Chinese models are not good enough for anything other than pair programming, which is just a very last-gen way of using agents.
This matches my experience with DeepSeek V4 Pro at Max reasoning, the preview version of the model kept regularly messing things up. About 30-60% of additional time to fix the output was needed.
On similar tasks, GLM 5.2 at Max reasoning screwed up maybe 20-30% of the time, while it still definitely made noticeable mistakes, they were far fewer in total and less egregious.
Kimi K3 at Max reasoning drops that value to below 10%, it's about as good as Opus or approaches Fable in some tasks. At High reasoning it also seems to be pretty close to Opus 4.8, not sure about the latest Opus model yet, but it's up there.
Only problem is that K3 is nowhere near as cheap as DeepSeek models, despite me personally liking the writing tone more (less Anthropic slop) and finding that it doesn't block my cybersecurity prompts, recently reproduced SQLi with a proof of context so I could justify fixing it.
I'd say as Chinese models get better, whatever moat Anthropic and OpenAI have dissipates. Currently the main things keeping me with Anthropic are their performance (tokens/second) and the fact that their visualization abilities within the app are pretty good.
I’m sure the next models will only get better, when they’re released. Also super curious about what Moonshot will achieve and the full DeepSeek V4 Pro release!
> Their moat, especially OpenAI is funding and hardware resources. They gain train models 10x as large and also serve at large scale. That's it.
I’ve seen how much slower Kimi K3 can be and that part seems correct, their own GPU production still has ways to go and export restrictions definitely limit what they can do.
Not sure about the size part, if Kimi K3 achieves SOTA performance at 2.8T parameters, western models being >2x that size would be insanely bad in regards to efficiency. I bet they’re all within the same order of magnitude and below 10T and won’t really have a reason to go even that high for the foreseeable future.
As investors will start squeezing them for profitability, I suspect focusing more on efficiency will be commonplace.
> Not sure about the size part, if Kimi K3 achieves SOTA performance at 2.8T parameters, western models being >2x that size would be insanely bad in regards to efficiency. I bet they’re all within the same order of magnitude and below 10T and won’t really have a reason to go even that high for the foreseeable future.
They're a lot larger e.g. Fable. It is insanely bad. Do you know how much more resources "Western" companies have? Most in China don't have random GPUs to "play with" like every "frontier lab" employee does.
> As investors will start squeezing them for profitability, I suspect focusing more on efficiency will be commonplace.
They're born lucky though. Efficiency is "free". The next generation hardware e.g. Nvidia claims Blackwell -> Rubin is 10x efficiency (verified by Neoclouds apparently).
> It’s very clear that vast majority of the business do not generate additional value from incremental intelligence gain from these models.
It's been 2 months since Fable was released to the general, man. Nobody knows what's going on inside of these companies except the people at the coal face.
> but it’s nothing like Fable. It’s still strongly human-in-the-loop which is fine,
It's funny to see that Anthopic shills have been saying the exact same thing for the past two years now (and it was OpenAI fans before). It's amazing to see that Claude 3 Sonnet was "great" but now that even Qwen 9B is better than this version of Sonnet DeepSeek V4 is still not good enough despite being stronger than Opus 4.7 was.
> Let’s say you could buy an LLM that gets things right 98% of the time. But there’s another LLM that’s 100x the price but gets things right 99.9% of the time
If you think Fable makes 20 times fewer mistakes than DS4 you're delusional. It doesn't even do 20 fewer mistake than Gemma 4…
The discussion started about cost-vs-usability, and then you brought in "but humans, the best LLMs around, cost much more" into this discussion to make it a not cost-vs-usability discussion. Do you not find it a bit disingenuous?
Not really, no. My point is that a $100/mo LLM subscription is generating $5000/mo in value, so even a negligible difference in performance wipes out the cost savings completely.
Same reason it makes sense to assign a team of humans that cost $100k/mo to a product that brings in $5M/mo, rather than one human with 5 Claude Max subs.
> You use Fable 5 right? If that’s good enough for you now, why wouldn’t a Chinese model that’s as good as Fable 5 but at 10% the cost be good enough in 6 months?
This is why I quite like Kimi K3 - close to the same performance (definitely like Opus, approaching Fable), noticeably cheaper, generally good enough for me to daily drive. Only problem is that their official provider (on the Vivace plan) feels kinda slow, I'd say close to 2x slower than Opus on Max reasoning on average (probably more relatable than Fable).
SOTA models are exceedingly good at self-correction in the right harness especially in domains where things can be proven mathematically. Most people are just deploying Claude Code or Codex with default settings and rub the genie lamp expecting exceptional results. Garbage in, garbage out.
There are a lot of tasks that are hard for organisations to run consistently but require some intelligence - monitoring logs and metrics for anomalies and security events, backup audits, audit processes in general, ensuring document quality and consistency, database advice and tuning, customer experience management, process optimisation - that are not "long horizon" in the classical sense of each step depending on the last, but are the result of consistency and attention over a long period of time and a large amount of data.
For this genre of task execution can run with limited horizon and is independent but would be too expensive to do with "us frontier tokens", I think for these, there is value in availability of cheaper tokens.
These are not 24 hours of inference with floating point errors accumulating; largely the system guards against errors compounding. Tool failures, compile failures, test failures, etc, push back against the model taking a wrong turn and force it to correct.
Yes it's much easier to have a smarter model that goes straight to the correct answer first, but it may not be necessary or economical. There's a minimum bar for the model where it understands problems and knows the right step to correct them, and above that newer models give diminishing returns.
> it's much easier to have a smarter model that goes straight to the correct answer first
That's basically ASI not AGI, if you agree humans are NGI (natural general intelligence) and make mistakes and wrong decisions in solutions all the time. Right steps with some wrong ones is acceptable though for AGI.
What you're overlooking is that any large task can, and should, be broken down into smaller individual components that can be reasoned about and tested in isolation. This is literally the whole basis for how we do programming. You don't need a model that can keep track on a gargantuan tasks all at once. You need a process for breaking problems down into manageable chunks, and then assembling them into a solution. This is a problem that can be solved by a harness through steering and and having a decent agentic loop.
White collar work will require SOTA models up until the point where said models can automate white collar work entirely. Then, we might finally see a meaningful commoditization crunch.
Decompiling / disassembling and annotating old software, making sure it can build cleanly back to the original binary, and then look for bugs or subtle issues.
Another one I did was a printer data stream translator from an obscure format to PostScript/PDF (or just PNGs), complete with cups support, etc so these old apps can easily be hooked up.
Flash is capable now of running long range defined-goal tasks like this.
I wondered this too. Also, the duration of the task depends on the quality of the prompt and the model used. I have a feeling a lot of these day-long tasks are bogged down by suboptimal tool use and on-demand python slop.
- You describe the breakdown in terms of time but it's more accurately a function of reasoning complexity.
- You seem to assume that no intermediate evaluation is possible.
- Often it is (e.g. the build breaks or tests start failing), allowing for course correction. There's definitely a cost to that but it can still be cost effective if the accuracy is "good enough" and the price difference significant.
- There are numerous tasks that don't require Fable or GPT5.6 level reasoning to improve efficiency by an order of magnitude.
For simple queries, we have reached the threshold since the beginning of the year, and models are good enough from every provider to make a meaningful difference between one another. (ChatGPT, Claude, Gemini, Grok, MuseSpark, Kimi, DeepSeek, GLM...)
The real unlock will be, and you can already see it with GPT-5.6 and Fable-5, to delegate complex enough tasks that will take more than 24 hours to get done and they will not lose track. I'm not talking about a loop, but the actual intelligence to recover from these compounding errors that accumulate in dumber models.
We're still a long way from the intelligence needed to let one of these agents go ahead and supervise multiple layers of sub-agents underneath to do complex orchestration. The future looks very promising and exciting. Imagine having the possibility of a Frontier model orchestrating as many sub-agents as needed that are running on cheaper models like DeepSeek.