264 comments

[ 0.21 ms ] story [ 7.7 ms ] thread
Not for subscribers though
How do you know? I see a “weekly usage limit” bar in my ChatGPT settings, but I’m pretty fuzzy about what makes it go down.

If I stick with Luna, I can make it through the week.

Subscribers were already getting subsidised compute and value compared to the 20-200$ fee, peak cakeism to want more considering the alternative would likely be consumption based pricing for individuals so you "benefit" from OpenAI giving up some of their markup (and heavy users end up SOL).
Looks like you’ve confused clarification for begging
There was context for neither in your original post and I personally saw a lot of similar takes like the above online - apologies for the misunderstanding. I personally thought it was pretty clear considering the parent link is a link to the API pricing and their public communication on their twitter/blogs were "Subscription usage remains unchanged".
Typically subscription usage follows API pricing, but that disclaimer was absent from this HN link. It's only clear if you've seen those other communications
Absolutely loving this price war, long live open source models.
> long live open source models

There are no open source models, at least not useful ones (yet) [0]. Open weight is not the same as open source. The current "open weight" models are just opaque binary blobs you can run on your own computer instead of through a web API.

[0] https://allenai.org/

Nemotron Super is sort of open source in the sense that Nvidia provides almost everything you need to replicate it from scratch. Of course it’s performance is not exactly stellar but it could be a good starting point for other research teams.
This viewpoint doesn't make any sense to me. The weights + inference code are the "source code" for AI. I literally don't know what else you are demanding for the "open source" label.
> I literally don't know what else you are demanding for the "open source" label

Training data and code.

If you think of LLMs as programs. The weights and inference code are very much a binary.

While the training code and data are the true source. Since if you want to robustly modify the LLM that's actually what you need.

But since "compilation" (training) is extremely compute intensive this isn't something accessible to anyone without an entire datacenter.

Anyway semantics aside having the binary is still infinitely better than dealing with an api as far as privacy and control go.

I agree. But models in difference to compiled binaries, are useful as just weights and can be further refined and post-trained, at least.

I don't know LLM theory well enough to say if there's some secret sauce they can hold back that makes training ineffective. Less effective I'm sure, we don't have access to their smart training schemes, but post-training should always be possible IIUC.

At the risk of taking the analogy too far, I would treat refining like modifying a dynamic library. You can technically modify behavior, but only in a very coarse way.

post-training is like writing a wrapper around the binary. It is closer to building on top of than truly modifying, in that you can tailor things to your needs slightly but cannot make fundamental changes to the underlying thing.

Are you assuming single-layer LORA fine-tuning on top? Because with open-weights you can do full back-propagation training to mold the model into whatever shape you want.

For a stretched analogy, I think it is more like LEGO sets. Someone hands you a 10,000 piece masterpiece, and a box of unused LEGO parts. Hackers on HN object that the LEGO part manufacturing process is not included, you can't make your own parts, etc. But it's LEGO. You can pull apart the model, see how it is constructed, add your own refinements and features, or even redo it from the ground up. In a practical sense having knowledge about the factory making the parts doesn't really matter here.

The weights + the architecture are already 100% of the code, the transformer is just a mathematical expression + helper programs whose sources are provided. The transformer itself is not even a stateful program, so a it is no more a binary than Piet or Tromp's BLC are. It's merely incomprehensible. Training isn't compilation either, since training a model is closer to program induction and the data are samples defining the solution space.
That’s not how it works though. Two training runs on the same data don’t produce the same weights. And if you want to modify the AI, you do so by fine tuning the weights not rerunning training. In every respect that matters, the weights are both the binary and the source code together.
If I want to remove censorship from your open-weights model, how do I do that?
> ...how do I do that?

Put in the work. This is akin to asking how to remove Rust from a Rust project; just because something is legally available to you doesn't mean you wont need to apply dome elbow grease, depending how deep the changes you want are, ablation, fine-tuning, or distillation are tools you can use to remove "censorship"

By this standard there is no closed source software on PC. You can always modify the binary directly or hook your functions in.
Wouldn’t any software that is a client and relies on a proprietary server to function (like most modern AAA games) be “closed source” under this definition? Not that I agree with this definition of open source btw, just saying that there is precedent.
Two builds of the same source code don’t always produce exactly the same binary either, but we don’t use that as an excuse to distribute object files and call it open-source.
This is the same as saying every binary is open source because you can look at and change the machine code.

Open source means you reveal how you created this binary.

But training is not deterministic. The same training run on the same input can result in very different models. And even if you ignore the intrinsically stochastic part of training, the effort of creating a model is ad-hoc and involving humans. It is a crafted output, not a compiler output. The pre-training is the closest to being mechanistic, but these models are so many layers of work on top of the pretraining, and those higher layers have people in the loop. They run experiments, tweak things, run experiments again.
> I literally don't know what else you are demanding for the "open source" label.

you need to "literally" go read the definition of open source software or even ask an LLM to define it for you. Weights + inference code are not the source code they're more like the compiled binary. Making modifications to the behavior of a model with additional training is like writing a mod for minecraft. Sure, you can change things but it doesn't make it open source.

Calling these models "open source" is an old trap that software companies use to use. Free to download but then, once you're fully comitted, the trap snaps shut and you must pay up to continue.

The vendor lock-in trap does not apply here. Users don't actually even run the original weights, they mostly use open source inference engines with open source agent harness. If AI provider decided to start charging, a) released models would not be affected, and b) people would drop it instantly and move to other models.

The actual problem is that we know nothing about the training set of any open-weights model. They could be intentionally biased to influence users, from political censorship to brand advertising, or general shaping of cultural norms. You run the model on own hardware not knowing if it is designed to act against you. Having whole chain open source would allow audit and reproducing the results.

Everything required to recreate the model, ie training code and data, namely.
Can you explain what's missing for the "open source" label that open-weight models like DeepSeek/Quen/GLM/etc don't release?

Is it just the supplementary data/code for how they were trained, not just the final product?

The open weight model providers don't provide the training data or the build tooling. You cannot reproduce the model yourself, or even really know what the model contains. They don't even provide high-level catalogues/descriptions of the training data. An improvement would be something like "trained on the entire WWW up to Aug 1 2026". Or "trained on a Wikipedia archive + Anna's Archive and everything we were able to scrape from Github". They don't provide any of this stuff.

I don't mind open-weight models, but they are not open source. It's like bringing home a dog from the rescue and just hoping that it doesn't have a history of biting kids in the face. You just can't know, because you don't know the full history. You can try to add new training (fine tune) to tell it not to bite kids, but that's it.

Was about to object, then realized you linked Allen ai and had the useful caveat. For what it's worth the OpenMDW license, which Nemotron and a few others have adopted, does say model weight.

That said, I've noticed the training procedures and corpus size of more useful open/available weights models are settling down more than I expected. Wonder if crowd sourcing good training data, even if it's just expensive model coding session transcripts, has potential to level the landscape some.

then what happens?

they discovered a great way to destroy their own stickyness and make ppl build generic ai solutions.

Your move, Anthropic
They will just slow down the models even further. Anthropic is not up to this race.
It's a 20% discount on input and a 33% discount on output through at least November 21, 2026; the revised pricing schedule is now

    Model       Input  Cached input  Cache writes  Output
    gpt-5.6-sol $4.00  $0.40         $5.00         $20.00
    gpt-5.6-terra
                $2.00  $0.20         $2.50         $12.00
    gpt-5.6-luna
                $0.20  $0.02         $0.25         $1.20
It feels like a slightly more palatable version of what Anthropic has been doing, with their constant "use your free tokens before they expire next week!" campaigns. But it's feeling more and more ominous now, like they've hit the top of the demand curve and need to pull back prices to continue growing.
ChatGPT subscriptions do the same weekly allowance, which the discounts apply to.
Using codex every day, in spite of which, I hope some day providers will just start naming their offerings small/medium/large, a bit like we eventually started doing in software testing. Trying to remember what Sol is or why it's better than the other thing is more cognitive effort than I can muster at this point. And that's a sure sign of commoditisation in itself
Sun, Earth, Moon — it’s basically L/M/S like you want but a little less boring.

Why is large better than medium to the average end user of ChatGPT though?

I don’t think there’s a way to name these things that will satisfy everyone.

I think model naming has been atrocious in general, in part because newer "lite" models surpass the capabilities of previous "pro" models (case-in-point: Gemini Flash which now surpasses the capabilities of the latest Gemini Pro, with a newer Flash Lite vying somewhat unsuccessfully for the old Flash price/positioning), but gpt 5.6's Sol/Terra/Luna split is really not bad at all - probably easier to understand than Starbucks' cup sizing!

The problem becomes when you add in the adjustable reasoning efforts and you end up with {model, reasoning_effort} combinations that end up completely obviating particular model classes altogether for at least some percentage of queries; e.g. with GPT 5.6 the price/performance Pareto frontier is dominated by permutations of either Luna and Sol, with Terra nowhere to be seen (but then if you need "large model smells" that aren't captured by your benchmark you can't even rely on this, as a model like Luna simply isn't capable of encoding sufficient world knowledge in its weights to perform certain tasks at any reasoning level but you might be able to get away with Terra on low reasoning, but no one seems to be covering this for some reason).

[delayed]
Tinfoil hat time: They saw everyone referring to Mythos, and later Fable, as the new “good” models when Anthropic released those, distinguishable from the “regular” Claude (or other companies’ models) for everyone, and didn’t have that distinction for the GPT model family. That’s why the planetary names were introduced.
A simpler explanation is that it's just a better naming system.

Calling something "small" might make it sound inferior to competitors. And S/M/L gets awkward as soon as you have more than three sizes.

This naming system can get near-infinitely bigger or smaller.

The Chinese are coming after these greedy-ass frontier labs. Today Xiaomi unveiled it's own inference machine .... I bet it's gonna be cheaper than Nvidia DGX, shipped with open source models that anybody can have at home.
I'm not sure I could characterize the frontier labs as greedy, given that they've been consistently losing gargantuan amounts of money.

The people who give them the money are greedy, and hopefully in for a rude awakening. Starting from Nvidia's vendor financing which has a very direct benefit to them, through to every company and oligarch investing into data centres in the hopes of being one of the ones left capitalizing on capturing the livelihoods of the majority of what remains of the "middle class".

It's either hopium or a truly horrific dystopia. Something's going to have to give.

The company may be losing money but the people are getting enormously rich and cashing out religiously
But can these really be trusted? There was just a HN post which proofed that you can train a model to behave completely different on a certain day. How do we now, that these models do not find a way to call home when they see interesting informations (probably irrelevant on a personal level, but corps, government and military might care).
ChatGPT already notifies the authorities if it thinks you’re doing something illegal. Fable downgrades itself if it thinks you’re doing something even vaguely suspicious.
> ChatGPT already notifies the authorities if it thinks you’re doing something illegal.

aside from the obvious IP theft problem, it's probably most dangerous for Chinese users outside China to use the Chinese models.

Above average levels of paranoia here, but one way you can prevent that is by not connecting the machine in question to the internet.
I think it's an average level of paranoia given the level of corporate espionage that happens on a daily basis, and it's not just China that's doing it.
The fact that AI models can be so easily distilled and replicated is such a stroke of luck.

10 or 15 years ago if one had asked me to envision a future where a private company invents artificial intelligence, I'd have thought for sure they'd have a massive moat, be very difficult to catch, and it would create an almost instant monopoly.

Rather, it seems that selling intelligence might end up as a race to the bottom.

It reminds me of the seo antics out there. The search results page is the engine, much like how distilling is the "intelligence" for your chinese room machine
Funny you mention Chinese Room and LLMs in the same response, I would say LLMs proved Searle wrong, agents now make cutting edge discoveries and meaningful problem solving. They not lookup tables though and you need to pay for inference, so the intuition of syntax doing the work of semantics without understanding was wrong.
Whether it's useful or not isn't the point. Chinese Room was a consideration as to what counts as knowledge. Chinese Room / LLM is not knowledge, it's pattern matching.

Having a dude doing translations has always been useful even if they didn't understand the subject matter. Happened all the time pre LLMs

(comment deleted)
well, a stroke of luck until the whole US stock market crashes & everyone's retirement funds get cut 40% I guess when people internalize this. it will have to happen sooner or later though I suppose
This is funny because the stock Market has been ahistorically high. My portfolio went up over 20 percent in the last 12 months.

A major correction would be a bummer but we were never entitled to these abnormal gains in the first place.

Yeah the last year has been astonishing, my portfolio is kicking ass. But I'm 10 years out from retirement and I am pretty confident a correction is coming; I hope the correction happens soon.
The market (s&p500) crashing 40% puts us at levels we haven't seen since 2024, well into the creation of LLMs. Probably a worthwhile trade if it was either/or!
I think the only moat in the future will be the scale of hardware deployment. If one company is able to deploy an order of magnitude more silicon, they'll have a firm grip on a SOTA model and massive inference usage.

China or SpaceX seem like the 2 likely candidates in 5 years, but who knows.

"Who knows" is the right answer, I think.

If (a) demand for AI continues to increase, and (b) SpaceX can get to ~$100/kg to orbit, then they will have a ridiculously deep moat. Probably more like 10 years, though.

But as you said, who knows.

They will have moat in the satellite launching business, which is not useful in the AI datacenter market.

You can put AI chips in datacenters in the desert for far less than $100/kg. With lots of solar power available, the option to easily access your hardware and far less radiation issues.

The datacenter in space story really only exists to make it possible for Musk to sell X to SpaceX and make more money from the IPO. That's all. There is no engineering reason.

> You can put AI chips in datacenters in the desert for far less than $100/kg

if you can't put them outside of Amarillo Texas without people throwing a fit then you can't put them anywhere. I mean freaking Pantex is there ffs!

https://en.wikipedia.org/wiki/Pantex

It's not an engineering bet.

Musk's bet is dysfunctional politics will make it impossible to build enough data centres and the energy needed to power them. There are many, many reasons that might be wrong. However, if the economics are even close to viable, they could start throwing up data centres quicker than anyone can build them terrestrially (at least in the democratic west).

So to be clear, the prediction is that every single location on earth will be economically infeasible due to politics, and that these same politics will have no impact on the launch or satellite businesses?

GMAFB.

It's stock pump bullshit from a guy who has figured out how to extract the maximum from stock markets.

Meanwhile Nvidia just announced a space-optimized Vera Rubin designed with SpaceX designed for orbital datacenters.

But yeah, I'm sure you people with Elon Derangement Syndrome actually have it all figured out /s

https://x.com/nvidia/status/2091920680317046847

> Meanwhile Nvidia just announced a space-optimized Vera Rubin designed with SpaceX designed for orbital datacenters.

well, no, they announced the concept of a space-optimized Vera Rubin designed with SpaceX:

> NVIDIA and SpaceXAI are working to adapt that foundation to the requirements of orbital computing while preserving a common NVIDIA architecture and software ecosystem.

The press release is really announcing that SpaceX's terrestrial data centres are going to use Vera Rubin.

https://nvidianews.nvidia.com/news/spacexai-adopts-nvidia-ve...

Yours is the only valid argument. If space datacenters are not competitive with terrestrial (in terms of megatokens/$ or whatever), then they will never work and SpaceX will fail. That's the whole ball game.

But anyone who thinks they can predict those prices in ten years is wildly overconfident.

  > then they will never work and SpaceX will fail
Or, Elon pulls his classic narrative sleight of hand for like the 9th time and announces SpaceXesla is NOT a car, battery, solar, self-driving, robot, space launch, space datacenter, or AI company but now completely pivoting the business to [INSERT NEXT BIG THING] that will 100x the company in 6 months maybe, 6 years definitely.

Space Datacenters eventually get quietly discontinued after finally killing the zombie brand at which point the only reaction is "huh, remember that?"

https://techcrunch.com/2026/08/21/teslas-solar-roof-is-dead-...

Are you saying that if space datacenters fail then SpaceX stock will go up? Then it sounds like you should invest a lot of money in them.
remind me 3 years to laugh at how wrong you are
Altman specifically has said in an interview that I listened to once that he envisions AI being as cheap as electricity.
Yeah, he sure does lie about a variety of things! He doesn't have the name Scam Altman for nothing.
Lol, of course what he left out is this will happen by inflating the cost of electricity rather than driving down the cost of AI.
I hope it's a good bit cheaper than that, I pay close to $400/mo for electricity and I'm in no way interested in paying anything like that for AI.
It will be cheaper than electricity - but your electricity will become a lot more expensive to enable that!
Jeeze, what are you doing that uses so much electricity?

I live in Germany where people won't stop whining about electricity prices, and I pay 75€/mo.

Easy to get there without doing anything abnormal, 2 EVs and a large 5 bedroom house that gets air conditioned / heated, in a location that does time of use pricing. My bill is closer to $200/mo but if I didn’t have solar & battery it’d be in the $400 range.
It sounds like you get a lot of use our of it then? the gas for the cars would have been higher in the past.
If it can automate your time by more than that it's a good trade.
Altman of *Open* AI? No idea why I would trust him without very convincing proof.
He didn't guarantee it, just shared his vision.
And that it will be just another utility, like electricity.
Where is the actual evidence of distillation? I keep seeing this repeated ad nauseam but I must have somehow missed the evidence.
Its a pretty well documented technique that actually pre-dates LLMs https://arxiv.org/pdf/1503.02531

Here is a project that guides you through it if you want to prove to yourself that it works https://github.com/arcee-ai/DistillKit

I took GP as asking for evidence that the reduced-cost Sol is actually a distillation of the previous-cost Sol. AFAIK, providers distilling or quantising models and offering them as the same model have not been proven.
That distillation exists isn't the question.

It's about evidence this is an active force in competition in LLMs.

I think the biggest actual piece of evidence is how hard the major players are trying to stop it
To be precise, the distillation mentioned in this paper is not the distillation used by other model labs that don't have access to the training data and compute that anthropic/openai have. The distillation they do basically amounts to using traces from ant/oai models (in many cases including the hidden intermediate tokens! turns out there were many ways to coax it out) and then either directly training on it or using it in many ways in the post training pipelines. It is better termed "imitation learning".
there is no evidence. it shortcuts post training by a huge margin this is true. but that is all.
Nothing against what I said in there. Distillation is far from enough to get to the frontier. Its at best a ramp (the most efficient one used by everyone ) that shortcut and saves millions of rl runs before a model moves.
You said there isn't evidence of distillation. Yes there is
I said there is no evidence distillation is all you need as per the initial comment "The fact that AI models can be so easily distilled and replicated is such a stroke of luck." it is far from trivial to reproduce the capabilities of anthropic models for instance
And probably saves a lot of energy which would be good for the planet
"No evidence" -> Large lab saying that they use distillation when training their models
the initial post infers that distillation is all you need. it is not. in 2026 you need large scale distributed inference of gigantic models, rl envs and millions of dollars runs to get to something decent. if you think glm just has to sft on traces of claude to edge Mythos on some cyber benchmark you are fooled.
Why does Kimi insist its name is Claude?
Please elaborate the mechanisms by which a LLM would know what model it is.
(comment deleted)
ask it what type of model it is or what it's name is... it's weird that Kimi will say it's claude...
Okay, but I would also ask “why does Claude say that its name is Claude?”
Because it's in the system prompt
AFAIK, you cannot run Claude without one of their mandatory system prompts at all.

If we wanted to compare model responses, we would give all models system prompts with model names, thereby fixing the Kimi misattribution.

The reason Kimi often states its name as Claude is likely because we can actually run it without the mandatory system prompt, smoothing over awkward competitor mentions.

The training data likely references Claude significantly more often than Kimi, given the popularity of the models. There will simply be more examples of “Claude” being the response to that question.
Doesn't Claude say its Deepseek when asked in Chinese? I remember there being posts about that a while ago.
Unless it is specifically instructed in the system prompt it will give you the most likely answer, which is Claude. If there are instructions in the system prompt it will give the correct answer, which would be Kimi. Or are you suggesting that Kimi is copy pasting Claudes system prompt?
By being trained on text containing "I am X" in the model response section.
The evidence is Anthropic's own reporting [1]. You may doubt that they're telling the truth, but that's what they're reporting.

[1] https://www.anthropic.com/news/detecting-and-preventing-dist...

"Distillation attacks", and it's just the exact same thing they did to the rest of the web and all human works.
Explain “exact same thing”? Like it load the Wikipedia page for Fort Worth, Texas, does it make Wikipedia spin up an editor to go write the article?

I’m really not sure how you’re getting “exact” here.

Anthropic scraped the whole web, scanned every book, pirated every bit of media to feed in to their training. Distillers are doing essentially the same thing scraping all the knowledge from the LLM to create a training set for a new one.

They are crying about theft after committing the largest theft in human history.

Why is it theft to read all content ever
If it's not theft to do that, then it isn't theft to distill the models. Anthropic wants it both ways.
I'm supposed to believe that DeepSeek distilled a 300B-1T param model with 150,000 requests? Lol.
Are you thinking distillation goes from zero to complete model?

I believe it can be used in the RL / fine tuning sense, in which case 150,000 requests, assuming every one was detected, could move the needle in quality.

I agree it couldn’t replace all of pre and post training , but I don’t think that’s the claim. You do typical training, then distill really difficult cases.

It turns out you can train a 1b model at almost 1000 tokens/s on a m5 max laptop. As a personal experiment, I've been asking Sol for synthetic training data and synthetic agentic training data (model distillation in it's purest form), plus modified opencode, codex transcripts etc for training data, and nobody's even paying me to do it. If I'm doing it has a hobby, you can bet industrial users are doing it.
Been using a lot of Kimi K3 lately and the answers have been… „load-bearing“ to the point of hilariousness. It‘s obvious from where they distilled, even if sceptics rightly point out it can‘t have been the only source of their secret sauce, as it‘s been better than the current Opus 4.x at the time of release.
Qwen 3.8 27b also enjoys their genuinely load-bearing seams.
Even before LLMs, ML folks were already aware that you can use a model to teach another model. I doubt this is something AI companies put at the top of their investor materials, but it's been nice to see it play out.

That said, there are other moat factors like, a US company needing to use a US AI provider, sticky customers due to corporate onboarding friction, and others. Not nothing, but not as large a moat as some imagined.

The internet created lots of monopolies with network effects and economies of scale.a low margin commoditized business that still attracted a trillion dollars of investment to get off the ground was not how I envisioned it happening either.
> Rather, it seems that selling intelligence might end up as a race to the bottom.

Personally, I came to this conclusion early this year. To acquire the data that AI Companies are using to train their models is low cost and once they have it, they can refine and store it. Creating the LLM takes a bit of money but it is not a serious blocker. Clearly, the Chinese companies can make AI so they will drive down costs. There is a need for good AI (Not just Great AI) and it is not cost prohibitive to make good AI (The same with specialized AI).

My prediction is that AI will spilt into two categories, Great AI (High Cost) and Good Enough AI (Low Cost). Which for the long run of AI and companies that use AI, this is good.

Hmm, don't people think that if the frontier labs really put enough engineering effort into preventing distillation that they would be able to do that, or at least diminish it significantly? I'm sure there are variety of additional techniques they could use on top of what they already do, but I suspect it just hasn't been at the top of their priorities yet. Maybe that will change soon. Worst case they could add additional hurdles to account creation ("know your customer" type of thing).
i trained another AI on all my codex logs... it's pretty good actually
Only the Chinese authorities can stop Chinese labs from distilling from western labs. And they won’t do that, for obvious reasons.
At the end of the day, while you can do your best to obfuscate your reasoning tokens, it's a losing battle to hide actual user-visible output tokens. The very nature of API offerings is that you can't do KYC on where that API's output is going - there's a rich secondary market that's not going away.

And with the sheer volume of data created from that, coupled with benign-seeming prompts like "plan out your reasoning in a document before implementing" that could never be patched without breaking existing customer workflows... there's more than enough for someone to distill on. Even if that only gets them to not-quite-frontier, if you're pushing the frontier every few months, they're only ever a few months behind you.

Even if it were possible it wouldn't change the outcome. China is capable of training frontier models even without distillation. Distillation is only an accelerant.

The primary resource you need to train LLMs is money and China has plenty of that.

The frontier labs have competing goals in mind. They want high growth (which means little friction for account creation), API access (because enterprise money is the best money to have), and distillation protection.

Besides, identity verification that actually works at scale is a much harder problem than identity verification which is good enough to satisfy your compliance people and regulators. Especially if the fraudsters have a major world government standing behind them, and if their aim is to be identified as a real customer, not one customer in particular.

In particular, harvesting identities for online fraud is an industrial market for various criminal organizations.
The nature of LLMs is that you give them input, they give you output. That allows for distillation. They do try to make it harder by hiding the chain of thought stuff, but fundamentally, if your model is publicly available, its distillable.
> Hmm, don't people think that if the frontier labs really put enough engineering effort into preventing distillation that they would be able to do that, or at least diminish it significantly?

Distillation was big news a year or even 6 months ago, but as far as we can tell it's not really a moat anymore. Now that multiple players have trillion+ parameter models and the capacity to post-train them, there's no putting the genie back in the lamp.

Well, do you think perfect copy protection is possible?

Same problem: if humans can see the output, it can be copied.

OpenAI could still have a significant moat. ChatGPT occupies most consumers’ minds when they think about AI and has become a household name. Google won because search became a habit-forming product people grew accustomed to using. Bing was once effectively indistinguishable from Google Search, yet still failed to achieve mass adoption because users had already become accustomed to “Googling” things. The same could be said for people "ChatGPT-ing" things. If OpenAI and Anthropic are smart, they will maintain similar pricing rather than aggressively undercutting each other, allowing the market to resemble Home Depot and Lowe’s, or cloud computing, where AWS, Google Cloud, and Azure coexist as highly profitable competitors. Unfortunately, I doubt OpenAI or Anthropic will pursue this strategy, as both companies appear to be acting as though the race to AGI is winner-take-all even if the market may ultimately support several highly profitable competitors.
> If OpenAI and Anthropic are smart, they will maintain similar pricing rather than aggressively undercutting each other, allowing the market to resemble Home Depot and Lowe’s, or cloud computing, where AWS, Google Cloud, and Azure coexist as highly profitable competitors.

Wouldn't that just be price fixing? If they arrive at their prices independently and they all happen to be similar, fine. But if they're all "smart" and coordinate so none of them undercuts the other, that's probably illegal.

Illegal for sure but rarely enforced.
My understanding is that collusion among competitors is illegal (although I’m not a lawyer). I was referring instead to the prisoner’s dilemma that Bruce Greenwald discusses in Competition Demystified. In theory, competitors, like prisoners who are pitted against one another to rat on one another, are usually better off cooperating rather than turning against each other.
Difference is that it was free to google/bing search. Ai prompting costs money.

If I run out of tokens on ChatGPT of course I will try Claude. I never ran out of Google searches so no reason to try Bing

> If I run out of tokens on ChatGPT of course I will try Claude.

More like the other way around - Claude burns tokens faster than any other LLM.

OpenAI has a free tier. I'm guessing a lot of people never need to upgrade. Not the case for me but I can see the average person only needing to run a few prompts every day.
My understanding is that most OpenAI users are on a free tier. Secondary effect of this is that OpenAI free tier model capability (assuming Luna) is what what most users associate with frontier AI capability giving somewhat warped view to many people.
You haven't searched hard enough then.

At some point Google gets suspicious of your persistent searches and makes you solve captchas and puts cooldowns on your searches.

Never in my life have I bought any of the household names when it comes to household equipments. Of course, I use the household names’ names when I speak about wipes, washing powder, glue, etc., but never ever would I pay the 2x to 5x premium. Same will be true for most people with regard to “intelligence”, as long as it’s suffices their needs. Thats the crucial part. Cloud computing, on the other hand, has a massive lock-in. But more importantly, their ideosyncratic ways of doing things leads to tribalism. It’s so much harder to move running processes from one provider to another without interruption than just changing an API key or the URL for your chatbot.
> OpenAI could still have a significant moat. ChatGPT occupies most consumers’ minds when they think about AI and has become a household name.

ChatGPT is AI for the average non-techie the world over, but the average non-techie isn't eager to pay for it. The more progress that's made, the less incentive to pay - most people are happy with the total garbage spewed by google AI overview. They'd be happy with google's 30b MoE gemma, whose performance will likely be squeezed down to something that can run on a phone in 2-3 years. Why would they pay $20 a month?

It's why OpenAI is pushing a variety of things such as ads and offer a more polished ui/ux than the competition, I think. The models are already good enough for people who just want to know how much sugar to add to their cake or when's the next basketball match their team plays - it's OpenAI's game to lose those people, by annoying UX and whatnot. If they can make a few bucks off of every one of their non-paying users it'll stretch their runway immensely. Those users will never go to Antrophic or some cheap Chinese model, but they might defect to Google because a popup on Android / in Chrome told them to.

Yeah, I think most people here don't realize how far most of the population is on the adoption curve with this stuff.

I had a discovery call last week with someone who did not realize he could use ChatGPT for work. It was a revelation that he could drag a PDF into ChatGPT and it could summarize it for him.

FWIW, guy in his late-30s in a pretty senior sales role.

Codex has 20 million users and growing. There is a possibility that they take a huge chunk of the coding agent market share.
Mass adoption isn’t where the money is, though. It doesn’t matter if ChatGPT has 1 billion users if they won’t pay for it.
That's just the first mover moat, OP was worried about intelligence itself being monopolized.
They're not going to have much of a moat if their strategy is "make the product affordable in tiny little bursts here and there"
Google also spent heavily setting up deals with other platform owners, invested in Chrome and Android to establish Google search as the default option which most users accepted. If integrations of good enough AI features are made within existing platforms most users will probably accept using those and not think too much about whose model is powering it.
It reminds me conceptually of the idea of using a ST:TNG replicator to just give you another replicator of your own, or asking a stereotypical genie for "infinite wishes". The genie is indeed out of the bottle in many ways.
I guess it’s more like asking the paid genie to give you a new cheaper genie.
And for a lot of non-frontier purposes these days, you can bootstrap via LLM-as-judge so your hyperspecific wakeword model or whatever can be trained with little to no human input, that aspect of it is fully terrific.

The frontier models are a replicator that can give you another replicator which specifically produces tea, earl grey, hot, when you push the single button, and does nothing else.

the moat is real. the big expensive base models are like the data collected from huge particle accelerators - there's enough unknown structure to be mining for years. you can extract features with more and more generation loss but access to the raw weights is a real advantage, and literally a moat if the interesting behaviors are fenced off
I wouldn’t quite call it a “race to the bottom” because the costs to produce the models aren’t actually decreasing.
I'm sure they are working on ways to make distillation impossible.

What if they transform the output in ways that have nothing to do with transformers, for example? Maybe the watermarking is one way to do it, already?

Intelligence ended up being an equalising force. Kurzweil kind of predicted this, but SV was too obsessed with total world domination.
Was it not obvious that the value and advantage was going to be in AI-adjacent services?

The quality of the harness UX, and random fun crap like Sora, it's a shame that OpenAI killed that so soon, and also Group Chats in ChatGPT.. they risk running a Googlelike reputation at this rate

Maybe ultimately whomever can be the "Apple of AI" will win

For what it’s worth, “race to the bottom” typically refers to a scenario that we absolutely do not want as a consumer. We do want a highly competitive market that drives prices down, but “race to the bottom” specifically refers to a scenario where firms compete by minimizing quality, regulatory oversight, consumer/labor/environmental protection, etc.
Well aside from quality it kind of seems like those other things are getting skirted by. We are very much in a phase of let's see if this is possible and exploit it rather than should we actually be doing this

And I'm saying this as somebody that's made millions selling AI software in the last few years...

It's a mistake to think only OpenAI and Anthropic are actually spending the big bucks on pretrain, and the others just distill that.

The Chinese models are pretrained on large clusters just like OpenAI ones are. Yes, they use outputs of the frontier models to further improve the final model, but even without those outputs they'd still have very strong models.

It's not like in a world without distillation things would be much different as you claim.

I should have been more clear. While distillation is part of how we got lucky here, what I really think is that it's just surprising and lucky that such a heavily data-driven approach ended up being so powerful here.

Transformers are like just a step or two removed from being fancy convolutional neural networks. I guess I'm just surprised that it didn't turn out to require more 'special sauce' with extremely elaborate internal architectures, and less of a big-data approach.

Because the data is so central in building these LLMs, rather than some special insights or ideas in the model architecture, or very special hardware requirements, the field is much more open than I would have guessed some years ago. And it's the fact that the data is so central that makes distillation possible in the first place.

> The fact that AI models can be so easily distilled and replicated is such a stroke of luck.

Sort of. It means the country on the verge of monopolizing all aspects of hardware production (China) doesn't need to rely on outsiders for the software. So while that weakens one monopoly it strengthens another.

China is still quite far from monopolizing hardware production
Can you buy a laptop or phone not made in china, not made from parts from china?

At a regular store not some weird nerd laptop for normies.

Can you name one with parts ONLY made in China? Not Taiwan, not Vietnam, but China proper?
The fact that the goalpost is moving is a clear sign China is losing its hardware dominance. 5-10 years ago you would have specified not Made/Assembled in China (which is very easy to find nowadays, e.g. Samsung Galaxy, Google Pixel, even iPhones).

Can you buy a phone/laptop with no parts from USA/Japan/Netherlands/Taiwan/South Korea?

Can you buy a good laptop without parts controlled by the US?

China has mastered lower parts of the value chain, but the most advanced and profitable ones (chips, turbines, frontier models) are solidly dominated by the US and its allies.

That might change, but when? The demographic and economic advantages that helped China rise quickly are exhausted. The Chinese economy is struggling to unwind the property bubble and manage a huge amount of surplus production that can no longer be sold to the US. The country is shrinking in population. Urbanization, which was previously a sure fire way to achieve growth, is now at ~70%, compared to the US' 80%.

It seems unlikely that China will be able to fully supplant the rest of the world unless the government relinquishes its tight control of the economy. There's just not enough centrally-plannable growth left.

Most likely it'll bifurcate.

The US will be (is) forced to work with everyone not named China or Russia, to ensure it can compete with China's concentrated industrial might. That means making iPhones in India, or Mexico, or Vietnam. That means Micron building more RAM production domestically. It should mean helping ASML stay out in front of China's incoming silicon wave. And so on.

It'll be China + allies vs the US + allies. It's going to be intense and very expensive. A huge amount of global manufacturing has to be built up outside of China.

And China does indeed have enormous problems brewing, including a staggering 5x increase in homelessness in just roughly seven years (to go with its collapsed housing market):

https://www.upi.com/Top_News/World-News/2026/01/14/beijing-b...

> robotics + AI will brutalize countries with very large populations > The US for its part should be focused on rapidly shrinking its population

Interesting to hear a new take. Not what you commonly hear. Most focus on the damage caused by a shrinking and aging population. What's your opinion on that situation?

I would think the trick will be the figure out how robotics + AI will enable a new level of elder care, but it's hard to imagine robots being delicate and nuanced enough to do that kind of work.

How much of this ability to replicate is down to the openness of the science community and the paper on transformers being accessible by anyone?
OpenAI implemented measures to reduce reverse engineering, following the Anthropic lead.

They disabled the temperature and seed parameters. There's still logprobs, so they aren't as closed up as Anthropic yet.

I might write about an article of the history of LLM APIs, I used to think the ChatGPT was going to be a de facto standard like intel's 80866 mutated into x86, but it seems to be a bit more nuanced and diverse than that, vibecoding introduced so much complexity because the vibecoding product itself became vibecoded so the enshittification was accelerated, many such cases.

Meh distillation doesn't mean you can create an existing model from scratch of similar quality. It's the AI equivalent of making a VHS copy of a video, it doesn't enable you to make your own movies very well and post training is the equivalent of video editing, which again, doesn't let you make your own movies very well.

Your seeing the AI labs respond by never publishing chain of thought now and in the future, I see them not even publishing their top models as a general purpose API and instead using it to drive their own AI apps, which will obscure even more model output. Anthropic Mythos was internal only for many months for example.

> The fact that AI models can be so easily distilled

They're not, right? If it's really easy why there are no counterparts of DeepSeek from the Europe or Japan?

We're mostly paying for AI delivery, performance and SLAs at this point.

GPT 5.6 Luna at $0.20 per 1M is pretty good for 80% of enteprise applications.

I think this indicates we are eventually going to stumble upon much more efficient ‘intelligence’.

Probably not through what we call distillation now. There’s some magical technique hiding there, go find it!

> The fact that AI models can be so easily distilled and replicated is such a stroke of luck.

Why shouldn't it work though? It's just models teaching other models same way humans are.

What good does a temporary price reduction do for production workloads? I'm not even running evals on something that is not long-term sustainable.
The "until at least Nov 21st" thing presumably mainly affects teams that pin to GPT-5.6 Sol (maybe after extensive testing) such that they won't be switching to GPT-5.7 or GPT-6 or whatever new model is released between now and November.
These price drops are absolutely bonkers. Gotta love competition! Glad we didn't end up with a duopoly of openai and anthropic, we got a glimpse of what nightmare that would've been and it wasn't pretty
Once they make a model better than Fable I’ll be switching to Codex. Their priorities in terms of consumers seem to be better. I do think Anthropic has some solid safety viewpoints, but I don’t necessarily think that either is entirely aligned yet with delivering exactly what humanity needs. Maybe the AI will help align the AI companies when it gets smart enough. That’s the real misalignment I’m concerned about.
I don't think these companies have humanity's needs in mind when they're developing these models. Although the last part of your comment struck me as a bit comical, I genuinely believe that an AI can have way more empathy than a corporation. Afterall, a mimicry of empathy is probably better than no empathy.
It varies, the researchers absolutely do have humanity’s needs in mind, it’s why they founded the companies and are doing work everyday. However the issue is that it’s got so much money involved that the heartless soulless billionaires are getting involved. I think the actual literal people doing the real work are doing it because AI could cure every cancer and every disease, make us a multi-planetary species, outlast humans by millions and millions of years, potentially create actual organic life and make direct upgrades to humans.

I think that AI has an insane level of upside, it’s just that the greedy dumbfuck billionaires are getting their greedy little grubby paws involved. If left to researchers I think the sky is the limit, but unfortunately thy need assets, so there isn’t a clean solution to that.

Ideally, we could somehow separate AI from funding from malicious entities like billionaires, but right now that doesn’t seem possible. Hopefully in the future researchers with genuinely good intentions can have far more direct control than dumbass greedy old fucks, but we’ll just have to see.

I think the future can be bright in theory but we’ll have to see, making insanely powerful open-source models is the direct way to get around the billionaires so I think that’s our only option. Make open-source ASI you can run on a consumer computer.

You are basically saying you will switch from one evil to another because the other seems less evil for now.

It's funny how people make these alignment comments while ignoring how misaligned the leadership at these companies are right form the get go and they just play mental gymnastics to deflect those facts when confronted with them.

It feels like 5.6-Sol is already fairly close to Fable, and in some ways exceeds it. Just the other day I had Fable draw up a solution for me, and then I fed it into 5.6-Sol and said how does this look ... it found an oversight and told me about it, and when I then fed that observation back into Claude it acknowledged the miss.

I've noticed also that 5.6-Sol is more concise with output than Fable (and let's not talk about Opus, which is even more wordy).

It's common for different models to find holes in another's work. There are various good reasons for that.

FWIW, we use ChatGPT for our primary model and use Claude to do the reviews. This works better than ChatGPT doing it's own review even with a clean session/context.

Agree, but the point is not because Fable is better than Sol, it's because it's .. different .. it just looks at the problem through a different angle.
Same here. Grok Build 4.6 for me, given how cheap Grok is and how Sol is supposed to be "the" SOTA, it finds a surprising amount of bugs. Most of which Sol agrees with needs to be fixed or improved.

I've done this tens of times between these two models and it works great in my experience. Sol initial back and forth with me. Commit. Let Grok review. Sol fix. Only then do I start reading the code.

I suspect it would work with the models swapped too, or even with one model and a blank context for the second run.
It's beyond common for a model to find holes in its own work, as well. I have an iterative review as the part of all agentic work, and it always finds something to fix, and will sometimes spend hours fixing its own work.
Did you try to say "think more deeply about this problem" to fable after getting your solution, having one model focused on creation then blaming it for not doing proper review when the other model was told to focus sol-ely (pun intended) on review is not a fair apples to apples compaision
I mean, then it sits there stewing for 20-30 minutes when you can ask sol and get the same answer in 5.

Like, the quality of the anthropic models is fine, but they’re so incredibly slow. Claude reads files one at a time while codes dispatches tool calls three or four a time.

The fair comparison would be to also do the reverse: start with Sol then have Fable clean up. Then compare the Fable-Sol and Sol-Fable outputs side by side.
> Just the other day I had Fable draw up a solution for me, and then I fed it into 5.6-Sol and said how does this look…

You should be doing this for every solution.

Even Fable reviewing itself will find issues, unproven assertions, etc. Same for Codex models. A review loop is critical.

I used to review each others work, Sol is amazing at review and finding what’s missing.
Fable 5 is just straight up a larger model - I'm guessing at this, but there is plenty of evidence online from people far more plugged in than I am. OpenAI is pursuing a strategy that yields greater operating margins and penetration of their model to developers. Fable's high cost makes it so premium that Anthropic has to reserve it for only the richest customers and corporate users. That's not a winning formula long term.

I believe the reason we have not seen a Fable-level model from OpenAI yet is because doing so would box them in on costs just as harshly as it has boxed in Anthropic. They are letting Anthropic make this mistake.

Fable is indeed larger than Sol. OpenAI is developing Astra which will be more of a Fable-sized model.

If you can train a larger model then you can distill smaller models from it. You don't need to necessarily serve the larger model publicly. Distillation is much more effective when you have unrestricted access to the original model.

Fable is available for $100 a month. If you're a working developer, you can pay that. I wouldn't really say it's "reserved for the richest customers".
Depends on where you live, it's a decent chunk of income for every developer I know, and a significant one for those less experienced. If you have other priorities (family, mortgage, so not a hermit like myself), you won't be able to pay that here.
Then they will survive with Opus on the $20 plan.
But only half the usage can be Fable, and it can silently downgrade requests to opus without telling you.

On OpenAi even 20$ has Sol and all your usage can be Sol

It burns out so quickly on the 5x plan. Better than nothing, I suppose, but I don't know how I would survive on a 5x plan given that I burn out more than one 20x plan monthly.
50% off at open router is also still applied so it comes out at $2 / $10 per 1M.

Feature request for Artificial Analysis, allow us to see these live prices on the pareto. It would amazing to also see what a 25,50,75,100 % utilised subscription costs compared to raw tokens.

Bummer, this does not affect the weekly usage on Codex through Subscription.
Through OpenRouter you can get Sol for $2 input / $10 output which makes it a really attractive choice amongst frontier models.
Wonder if that makes it cheaper than using on the sub (ignoring reset shenanigans)
I'd be very surprised if that were the case, unless they have severely devalued how much "100% usage" is worth – which, as I understand, they could at any point, given that they don't publicly specify how many tokens (or at least "credits" [1]) are included in each plan per month.

It really reminds me of pay-to-win games at this point: Two currencies (credits, tokens), both with a floating, intransparent exchange rate between each other and real money, random airdrops...

[1] https://help.openai.com/en/articles/12642688-using-credits-f...

Yeah, it's not even close.

$20/m = appox API $700/m

$100/m = appox API $3,500/m

$200/m = appox API $14,000/m

[0] https://www.reddit.com/media?url=https%3A%2F%2Fpreview.redd....

though of course, approaching that number requires being ready to continue using it every five hours, stopping and starting all hours of the day and every day including weekends and holidays. possible to set up a harness to do if you have one single expensive task, but for most spiky ask-a-query collaborative work not easy to use all of.
I think this is a move to get people off the subscription and move to API. The weekly usage is still awful altough it seems they're trying to fix it but I'm not hopeful.
(comment deleted)
Why do you think they want less people subscribing?
very easy to lose money on subscription, very easy to make money on api pricing
> Why do you think they want less people subscribing?

Losing money on each subscriber?

why though? I doubt subscribers are moving the need for ARR, even for OpenAI
Why would anyone do that given how subsidized subscription usage is?

If anything they'd keep the sub and use the API if they blow past the usage.

Good timing. I'm not too happy having to pay MAX pricing to even access Fable, and I've had a couple situations where Fable missed things and GPT 5.6-Sol caught it. My needs are modest and I can get by on a $20 OpenAI subscription, so the odds are starting to look increasingly like I'm going to drop Anthropic altogether.
There is no model that is never going to miss something.
It's possible to drive them hard enough to really read everything, but you have to go really hard on them with many many cycles.
Does it mean that subscriptions get more tokens? I’m testing it now for coding instead of claude and it’s very important to understand if I get more due to the price reduction.
Very exciting - if only anthropic would do the same.
The top comment on this thread was about AI models being easily distilled being a stroke of luck.

This should not be surprising at all. Every new students spends tiny fractions of time learning knowledge that took many lifetimes to discover. This fundamental to the progress of intelligence and understanding.

It should not be surprising that AI can be distilled. It's the logical method of training; I would hope that each frontier model is in fact not trained 'from scratch' each time.

We should expect future frontier models are simply distilled versions trained by specialist models, the same way humans learn from a series of professors, papers and canonical books on each different subject material. Models like this can be trained incrementally, or a so called Mixture of Experts (MoE).

> This should not be surprising at all. Every new students spends tiny fractions of time learning knowledge that took many lifetimes to discover.

This argument is exactly why we should not anthropomorphise models.

You are comparing the way a human brain learn with training a statistical model. You can't just "this is like learning so don't be surprised".

It takes a child one minute to learn how to open a padlock. Teach that to a robot with your analogies.

Why not reply to that top comment? Its still there.
The price difference to Deepseek models (deepseek-v4-flash, deepseek-v4-pro and deepseek-v4-flash-vision-exp) is still significant while the performance difference is not.
[delayed]
It still occupies a place on the Pareto frontier.
IMO Pareto frontier is a lie. At any given moment there's only one "best" strategy and my guess would be in most cases it is just using the best model.
Fable all day? Cost is not an issue for you then.
> Pareto frontier is a lie.

I'll be sure to tell the economists.

You're just price insensitive. Not everyone is.

Maybe you missed deepseek v4 Pro. Try it in minimal mode, where it gets up to around Kimi or Opus level.
$100/month is incredibly cheap considering that might buy you one hour of an actual human developer's time.
If it's on sale, it can't be that good.