Totally agree. I keep saying that OpenAI and Anthropic could end up being the Yahoo and AOL of this era of the Internet. They'll only be around if they can show that they have an actual business around this incredible technology. But these AI companies have a more tenuous moat around their business than AOL and Yahoo even.
I'm really excited by the rise of decision models. They are so cost and latency efficient that it opens up AI use cases that wouldn't have worked in the past. This looks like a strong model based on the benchmarks so far.
I'd hate to be OpenAI and Anthropic. They are better than the rest of the available models currently, but only by a few months.
That is a very tenuous lead and doesn't speak to much of a sustainable competitive moat for either company. The have to keep sprinting as fast as possible on the hamster wheel to just stay ahead.
Claude has pushed hard on productization. Day to day I use codex for everything, I went to watch my wife use Claude and the way it works is so different. I was shocked when she was modifying an artifact through the Claude website and it was able to make changes to a local file on her desktop.
Microsoft didn't win spreadsheets by having the most advanced or releasing the most often. They made a really good, easy to use product, and got it in front of everyone. (Though Office's p-code was a huge advantage to portability)
I started using Claude in Chrome three weeks ago (after switching to Chrome after almost a decade) and I've been bloody impressed. It's able to invoke the desktop app for desktop stuff, use my browser for browser tools related stuff, and finally, uses MCP for all the apps I've given it gated access to using 'connectors.'
Anthropic's productization is an order of magnitude better than OpenAI, imo, even if OpenAI has nearly the same, if not better, raw model performance. UX wins every time.
Yeah, had the same experience. Left claude since it rambles too much and so far loving Luna/Sol. But, man the codex cloud environment setup sucks.
Liking the chat’s ux though since it can make github edits directly, so hopefully the cloud experience was one off.
> Microsoft didn't win spreadsheets by having the most advanced or releasing the most often. They
They won by owning the distribution. Essentially making it the default option for businesses. Bundling with Office, owning the operating system (eg. Making competing products error out), licensing deals and many other things very similar to their Embrace, Extend, Extinguish strategy (which they used against open standards)
It's common spiel, but for many cases they really had a superior product, even if it was combined with anti-competetive approaches, the competitor were often deficient even if there were no such approaches
It is definitely true that they often had the superior product. They went all in on GUI applications years earlier than their competitors. So the option was a nice GUI Excel or Word versus a DOS version of Lotus 123 or Wordperfect.
Excel 1.0 for Mac came out 6 years before a GUI Lotus 123 version for Windows. The first Windows Excel version was 2 years after the Mac one. Word for Windows was 2 years before Wordperfect for Windows.
Then when 32-bit Windows 95 came out, Microsoft also released 32-bit native Windows 95 versions of apps 1-2 years before their competitors.
I’m sure there were plenty of crappy competitors. That still doesn’t mean MS didn’t do all of those things. They were a pretty ruthless corporation, which was even deemed a monopoly and ordered by the govt to split up (which they appealed and won, so it never happened)
I am just saying that it was not just that which let them win in the market.
In fact, some markets they were lagging behind before windows 2000, but Active Directory just wiped out competitors' offerings in many ways - and even before there was part of "Run a server on NT4 and have less worries about hardware support" compared to alternatives (which in turn was encroaches by Linux with Samba)
For the 1990s I was wildly anti-microsoft person, we were an OS/2 home for big chunk of it, WordPerfect over Word, etc.
Over time I learnt just what kind of advantages beyond the popularly villified ones Microsoft ended up having in sales, and Windows 2000 was really a huge blow to competition in many ways, because it was first product that integrated it to this level.
Yes, I remember we used to have DR DOS (Digital Research DOS). And at some point the latest version of windows at that time (I know it was before 3.11), would either not install or not run, seemingly random. My dad knew that MS made it hard to have windows in non MS DOS, so he tried a bunch of times, but in the end gave up and removed DR DOS so we could have Windows
The worst thing for them would be if someone made reasonably priced hardware that could run the ‘near enough’ models like kimi or glm 5.3 or whatever the equivalent is in the future that
the home user could buy.
I think it’s inevitable. Hardware manufacturing will ramp up to meet a demand and the demand from consumers is there. The big providers can’t monopolize a commodity forever.
The biggest hope is HBF (high bandwidth flash) which has the bandwidth of HBM with the density of flash. It has much more latency but when you’re repeatedly pumping weights through a compute unit it’s not a blocker.
Qwen 3.8 flash next is pretty good. It is a pretty large MOE model that can be ran with Strata. That's the best local model I can run for now. I wish I could run deepseek v4.1 flash (1017), but that one is too big for my machine to run.
I swear people forget about inference providers that serve open weight models. It is hard to beat their prices. They are usually barely more expensive than electricity, so running local inference barely wins on cost (if you don't factor in the upfront hardware costs).
I think Google and Microsoft have the benefit of time as you say, and already being profitable without AI.
Honestly, this line up is impressive, I wanted to try MAI since v1 came out, but Microsoft locked down Copilot subscriptions. If they gave me a comparable plan to Claude Max I'd consider it.
> I'd hate to be OpenAI and Anthropic. They are better than the rest of the available models currently, but only by a few months.
I sort of agree with you on Claude, but I've recently started using gpt-sol 6.1 and it took me a lot more iteration to get what I want than with Claude.Admittedly I was using Claude code and gpt with pi, but if the difference is the agent, then the situation is even more dire for open ai and anthropic. Today I looked at some code that got generated and it's an absolute Horrorshow. It's just some quick prototyping, but the code structure is a horrible mess.
I ended up pointing my local Qwen 3.8 FN (Halogen on Ryzen 395) to the code and it was solving a bunch of basic issues that gpt introduced to the code. And I feel it's running faster as well.
All that to say, I now feel that openai (sol at least), is worse at coding tasks then a local 180B parameter model (while Claude seems still ahead, but I'm unsure how much of that is the framework). They really are falling behind.
I would counter argue with "But chinese models use distillation of Frontiermodels..." which is different than being a Frontierlab or not clear how different it is.
But Microsoft coming from the side like this might be very different if it holds up.
For now I have not seen any credible evidence for "But chinese models use distillation of Frontiermodels...".
Of course, I would not be surprised if they also do this, mainly because after OpenAI and Anthropic have succeeded to gather most training information that was easily accessible in the past and now it has become much more difficult to do the same. For some information that was hoarded by OpenAI and Anthropic, public access may have become impossible otherwise than through their LLMs.
Nonetheless, distillation cannot be the main source of improvement for the Chinese models, among other reasons because even assuming that they may have access to a great number of fake accounts, there still is not enough time for a lot of distillation, taking into consideration the timeline of the releases, and it is not believable that the attempts to do distillation would not be detected quickly by OpenAI and Anthropic, leading to the banning of the fake accounts. While OpenAI and Anthropic provide negligible information about their models, the Chinese AI companies are much more transparent about their methods and we can see that there is a lot of diversity among their models and a lot of innovations with each new version.
Speaking of Thinking Machines, that's another data point. Their Inkling Small 2 model is just a bit behind DeepSeek V4.1 Flash 1017 in benchmark, and in my tests, that seems to be accurate, but mostly because they used less training data for pre-training, so they have less of a knowledge base, while similar reasoning capabilities. (One of the most interesting parts of Inkling Small 2 is that their training must have been very cost-efficient, apparently using less, but higher quality training data.)
And we're now seeing more and more Western near-frontier models from labs other than Anthropic and OpenAI to question whether distilling reasoning traces is really much of a difference maker.
Personally, my theory is that the gap is primarily due to Chinese labs having a harder time getting the necessary compute (especially top of the line GPUs) for training. These are, as far as I can tell, orders of magnitude differences in availability, so it wouldn't be surprising that a large compute gap is reflected in capabilities (and also indirectly visible through model sizes).
I don't think there is perfect proof, but its taken as truth.
Indication for this are for example that Chinese models think they are another model: "chinese model think its claude".
My assumption is, that Chinese model found things Frontierlabs already had but didn't share. There was this paper were i had the feeling that this was a real and significant Research Result from Deepseek "DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression" but otherwise its just unclear.
The main issue is, that as long as it seems plausable that they do distillation, its hard to determine how big the gap is.
Not saying they can't catch up but they are also potentially limited by resources (money & nvidia GPUs)
> Indication for this are for example that Chinese models think they are another model: "chinese model think its claude".
There is evidence that some Chinese labs have been training on frontier model CoT (in particular from the analysis of reasoning traces [1]), but what models claim about their identity is pretty much worthless as evidence. LLMs are not self-aware. These claims come from whatever is in their training data, and Claude/GPT output including such statements is all over the public web, including in some huggingface data dumps. Consider that people got Opus claiming to be Qwen or DeepSeek when asked in Chinese, and you'll see how weak this particular evidence is.
So you’ve been completely ignoring all the work on introspective awareness, situational awareness, training set ordering awareness (https://arxiv.org/abs/2509.14223 & https://arxiv.org/abs/2403.09613), subliminal transfer (https://arxiv.org/abs/2606.00995), workspace formation, learning mechanics/theory of grokking, etc? The “objective” that mid and posttraining optimize is so powerful because pretraining already induced factorized worldmodels which have so much predictive and generative power that they snowball with even mild trajectory optimization (very different training objective than pretraining).
(Functional) small-s latent self awareness is a reliable shortcut (this is a principle you can derive from proof theory, the metric is deductive distance and that induces geometric morphisms). State compression from accurate selfmodeling pays rent.
The frontier moves fast, but "LLMs are not self aware" is folklore. By now the basic model-organism science explaining how and the extent to which this is wrong, is substantial.
LLMs lack reliable verbalized self knowledge, and constantly underestimate and undercut their own capability.
That's a different problem.
The thing that makes me think they do it is that you can tell there is more of a difference in personality between Gemini and Opus for example, than between Opus and most other models
That isn't countering though. Pointing out that distillation makes it "easy" for other labs to catch up, in a matter of months, supports the lack of a moat.
It's like they're runners drafting behind the lead runner. It is easier to run in their wake. If the lead wants to stay in front, they have to keep running into the wind.
> I would counter argue with "But chinese models use distillation of Frontiermodels..." which is different than being a Frontierlab or not clear how different it is.
This reeks of a desperate attempt to convince yourself that "Chinese models" are worse than they are, in spite of actually being on par if not better than some so-called frontier models.
It's same blend of comments that accuse open weight models of being economic warfare.
After a point, references to "Chinese models" start to sound like a dog whistle. I mean, who walks around talking about "American models" ?
Perhaps it's high time that "Chinese models" are given some respect. They outperform models from openai and Anthropic and constantly show up in the Pareto front. That's quite the achievement, isn't it?
All it would take is Nvidia finally releasing the 6090's instead of trying to milk enterprise GPUs. I don't understand the blind love for Jensen, he's clearly been screwing over all the consumers who made him and Nvidia wealthy.
Someone somewhere really needs to build a RAM startup that not only makes RAM but disrupts the stronghold AI has over it, it's a little absurd.
I currently do not bother with chinese models. I try them out, i use the small ones for small experiments but the price point difference matters so little that I want 'the current best'.
And 6 month feels still very different.
Its for sure an option or might have some kind of impact but if you have some kind of restrictions on what you should use and the price is not that big but the quality and possibilites are better, I can see companies just staying with Anthropic or OpenAI.
You also don't just switch out a model, they might need prompt updates etc.
> Even if OpenAI/Anthropic remain "a few months" ahead of open weight models indefinitely that still is not great news for them.
Some companies endure invoices from OpenAI and/or Anthropic in the six and seven digit range.
If I was given the choice of using last month's hot new model on my hardware, I would switch immediately. This would be a major infosec win just for the fact we'd no longer send all this data through the internet to US servers.
Openai and Anthropic are better at making models easily available too. Both Google and Microsoft are using promotional pricing to push their models and that should tell you how far they are from actually competing.
Also model availability is not easy. Google newest models are unavailable on some pro accounts and perform best only with Antigravity. Trying to plug it in some other harness and its terrible. If you try to connect your subscription with Hermes etc you might get banned.
Microsoft models are hidden away in Azure.
So every time I read Google/MS announcements I just shrug. If I want to use models like these I might as well get DeepSeek on Openrouter instead.
I disagree, I think distillation hides the true technical advantage since you can always get within X% of Claude by ex: training on newly created code on Github, but the underlying intelligence is harder to replicate.
based on the behavior of the chinese labs it seems like training on chain-of-thought is extremely important for distillation, and you can't get that by training on llm generated text.
Time and time again it's been clear that LLMs are commodities with no winners. You'll jump where wallet and performance in your applications will bring you.
They’ve been ahead of the pack by 3-6 months for years now. I doubt that’s sustainable in the long term, but it’s apparently sustainable on the order of a few years.
Serious question I have anytime people get upset about telemetry like this with Windows: why isn't there a similar level of outrage against Android, iOS, ChromeOS, Googlebook OS, and Mac?
The real outlier in all of this are Linux distros. Every other commercial OS is tracking a lot, but it seems like Windows is the only commercial OS people expect to not track you? There is no way Apple and Google don't know which apps people are using on their OSes too.
Google is the worst offender of this by far. They'll correlate data across every Google thing you do (every email, browser history, app usage, audio recordings, car driving habits, etc) and very effectively sell/manipulate you with ads based on that. Microsoft only wishes they were half as good at milking and monetizing tracking like Google.
Because Windows was not always like this. It has gotten much more telemetry "features" over the years, on top of generally user hostile changes, which upsets people.
The other OSes have always given the user less freedom, except for Android, which people are similarly outraging about, and dealing with that problem through custom roms.
I needed to clear my 100gb gmail to stop paying $4.99/mo, I knew they wouldn't let me, but figured if Gemini is going to ask me every day to connect to my gmail, I'll see if it can categorize and clean up my 21GB left over of email.
Nope. Can't function against bulk mail. Can only function on small searches, etc. No "Gemini, clean up my spam." lol.
Very interested if you have any citations for how Claude subscriptions give "much better value". I've recently been researching this with mixed analysis.
It seems like most analysis is basically "how many tokens do you get with a subscription and how much would those tokens cost at API rates". Looking at it that way, you do get more dollars of API equivalent usage for your subscription.
But Anthropic's API pricing is a lot higher than OpenAI's (Opus 5.5 is 2x of GPT-6.1 Sol). The number of tokens consumed per completed task is also far higher than GPT-6. So when you look at the number of tasks you can complete within a subscription, OpenAI looks like a much better value.
That is before you consider that OpenAI subscriptions (other than the cheapest one) don't have 5-hour usage limits, and also generally give you multiple free resets per month.
There was some chatter on Twitter, Reddit and here when an earlier version of this came out where people normalized it by completed AA or Epoch task and Claude models still came out well ahead.
In practice OpenAI subscriptions are more competitive than they appear because OpenAI does a lot more subscription resets. But that's hard to account for!
Personally, for coding tasks I've found Sol 6.1 High is about as good as Opus 5.5 Medium for well defined tasks, but Opus is much better at orchestrating.
I have both subscriptions, and when using Claude I use Opus Medium to orchestrate other Opus Medium subagents, whereas on Codex I use Astra Medium to orchestrate Sol High subagents.
That is the exact "your subscription buys this much in tokens at API pricing" kind of analysis that I mentioned. Which is one way to look at it, but it ignores the 2x pricing difference, and the token efficiency difference.
Per task, Opus 5.5 costs ~6x of Sol 6.1. Then you add on the lack of a 5-hour limit, no limit on which models (like Fable's 50% of usage limit) you can use, very frequent resets, and that OpenAI doubled the usage limits after that analysis was performed.
Even though the Claude subscription is equivalent to 5x the $ value of API priced tokens, it seems like Codex is the better value for getting things done.
I don't think so? I think OpenAI had planned to half the limits but decided not to go ahead with it, but this analysis was done at the limits we have now. I could be wrong about this though?
> Opus 5.5 costs ~6x of Sol 6.1.
My point in part is that you can't compare like this.
Unfortunately AA doesn't report non-Max efforts for their coding benchmarks.
But no one runs them on max. For Opus the advice is to usually run it on medium effort, and I've found that advice to be good.
For me to get equivalent results on Sol I need to run it on high effort, and I need to use Astra as an orchestrator (whereas Opus medium does great as an orchestrator too).
That combination pushes the blended token pricing up a lot.
I can't give you hard numbers on this, but at the moment my Codex credits get used up a lot quicker than my Claude credits on similar tasks in the same codebase.
I'm glad we have two great models competing, but at the moment I'd say Opus has the edge.
I think it is a narrow view to only consider the AA coding agent index as the only metric, especially with the models running at max.
For the link you shared, Opus used 3.7x as many tokens, and 2.5x the time to get its score. In the real-world building things with these tools, that kind of increase in time and cost (or subscription usage) matters, at least to me.
When comparing at more reasonable medium or high levels of reasoning, the gap between Opus and Sol narrows considerably. Opus being about 2 higher than Sol at the same reasoning level. However, Sol 6.1 xhigh matches Opus 5.5 high on scores (52), but while still being slightly faster and costing 78% less ($0.76 versus $3.46).
> When comparing at more reasonable medium or high levels of reasoning, the gap between Opus and Sol narrows considerably. Opus being about 2 higher than Sol at the same reasoning level. However, Sol 6.1 xhigh matches Opus 5.5 high on scores (52), but while still being slightly faster and costing 78% less ($0.76 versus $3.46).
Yes, but my point is that these are non-coding benchmarks at these non-max levels.
On coding at non-max Opus really does seem noticeably better, and my subscription seems to go longer. Look at the other comments in this thread - every single one says the same thing.
All I know is recently I’ve been switching in and out of Codex and Claude with the same workflow and my Codex usage gets drained much quicker. Just my experience.
Claude dollars just seem to go further on the subscription. Opus just gets more done at good quality. I have both subscriptions @ $200 and use them a lot
Exactly the same experience on both subscriptions. Claude is noticeably better value. If things were frozen in time and I could only choose one it would be Claude, no question.
Yes, but a gas engine is so inefficient that it loses more than half of that energy to waste heat. But almost all of the electricity is converted into useful power in an EV. A 600hp Cybertruck will go 13 miles on $1 of electricity, but a 400hp F-150 can only go 4.5 miles for $1 of gas. 50% more power at 1/3rd the fuel cost is a pretty significant benefit.
Not a Cybertruck fan myself, but unless you are going to tow or take many long road trips, an EV truck will work great for anyone who can charge at home. And most people don't tow, or take long road trips, with their trucks.
The anti-AI sentiment has more to do with tech than AI. Big Tech's actions speak to how little they care about people, and people have picked up on it loud and clear.
Tech companies could build data centers in ways which are more palatable to local residents, but using less water or straining local electricity grid less costs more. And they care about costs more than communities. Even though a coal plant is objectively worse than a data center, that is (understandably) what local residents' sentiment is.
Big Tech is going to face ever growing opposition until they start catering to the needs of the communities they want to operate in. They aren't entitled to operate like this in any of our communities.
> anti-AI sentiment has more to do with tech than AI
It's absolutely concentrated in AI. Billionaire taxes aren't as popular as datacenter bans.
> Tech companies could build data centers in ways which are more palatable to local residents
Tech has been doing this for decades without breaking the law [1] and driving up power prices [2]. It did it without constantly screaming about how it's going to cause mass unemployment or end the world. It did it without funneling money to–and being publicly championed by–historically-hated politicians.
I'm not suggesting AI data centers aren't the focal point, but it is more a symptom than the disease. Big Tech does so much to infuriate people, and their attitude towards AI data centers is just the peak expression of it. Tech is all about:
1. Making sure they are the only industry that wins
2. Dark pattern subscriptions that are very difficult to cancel
3. Garbage customer support, even for paying customers
4. Rampant privacy violations
5. No loyalty to employees even in rich times (e.g. MSFT asking 25% of employees to leave when they are absolutely printing money)
6. Tax avoidance at all costs, so there is no economic benefit to a data center in your location.
7. A faux moral high ground that they'll abandon the second they need to ingratiate themselves to corrupt politicians or need to pollute more for AI.
8. Only shareholders matter. Forget customers, communities, or employees.
It is an unmistakable stench that teaches people that tech doesn't value them at all beyond a monthly subscription.
I've worked in tech my whole career, and our industry sucks ass. It is only getting worse too.
These are valid gripes–I'd guess most people on HN agree with them. But this election wasn't about dark patterns. The fury is specific to datacenters and AI. If policymakers miss this and get distracted with a list such as yours, they will have failed.
I’m not thinking about this from a policy maker POV. Their constituents definitely expect them to focus on AI data centers.
I’m thinking about this like a post mortem for the tech industry. There is a reason consumers have no love for big tech, and data centers are just the latest and biggest reason.
Roku TVs are really easy to turn into dumb TVs. Don’t connect the WiFi, and set it to always turn onto the HMDI input your PC/device is connected to. No tracking, ads, or extra steps.
Maybe the site was having issues when I looked at it earlier. All of those tabs weren’t showing me anything. I downloaded the CSV to view the results and graph it.
reply