Comment by johnfahey
5 days ago
OpenAI won a lot of good favor for the generous Codex subscription and the efficiency of their models, but now that many people have switched over from Claude, they think they can leverage their position to peddle a stream of unnecessary products, and crack down on the generous limits[1] that brought everyone to Codex in the first place.
Anthropic did the same thing. Earlier this year, Claude subs and Claude Code took off because of the subscription's incredible capability and value, then once they gained enough users, they started focusing on unnecessary products no one asked for (see Claude in Slack), and eventually lost their lead. After losing a bunch of customers to Codex subs they realized their mistake, and now they're shipping again.
AI companies are bad at making software; they are good at making AI models. And that's about it.
Nerds suck. They are smelly, have bad posture, manners, never leave their rooms and are generally unappealing.
They also have a nasty habit of being aware of nefarious practices, will resist all attempts to and harvest their data, or lock them into your service, and will drop you if your competitor makes a 3% better product, and will reject every upsell for actually profitable services.
Plus there is only so many of them.
A wave of nausea hit when I first saw the warm, fuzzy, cutesy, kawaii, Teletubby-like avatar that Meta gave its Muse agent.
And now OpenAI has done the same thing with their fuzzy, friendly, colorful dots.
I am physically sick.
im crying and throwing up everywhere. everything i use must be brutalist.
even seeing the linux penguin makes me want to punch my monitor and commit sudoku
18 replies →
I miss o3 in that regard. if GPT-4o led people into virtual romance and over-validation, o3 gave me the same sort of "madness" but with work.
when I tried GPT-5, I was sad mostly because GPT-5 had some added wordiness. fluff, if you will. o3, though, is a black hole you can talk to, and it only gave information back if there was something to give back. that kind of vibe is my dream coworker
3 replies →
The best part about Muse is they think that the Enterprise is going to start buying things from Meta that looks like it belongs in a preschool. Muse and Ray-Ban surveillance glasses, Zuck must live in one hell of an echo chamber!
3 replies →
Yes, it is hard, psychologically, to grant trust to some silly perverse little thing I want crush like a surreal cartoonish cockroach.
You wouldn’t survive 5 minutes in Japan
2 replies →
It's not for us. It's for people who didn't see value in these products. And for the press. Doesn't make it good, but it can still be effective.
It’s really remarkable isn’t it? These are the very same people who are cosplaying Oppenheimer and comparing these products to the bomb.
Youre probably not the target audience, but at the same time, who exactly is?
1 reply →
Physically cringed at this comment
Absolutely ridiculous comment can't believe anyone would actually post this seriously.
I remember a while ago the discussion kinda steered from "The leading LLM provider will be the one with the better model" to "...will be the one with more user history".
Tools like OpenClaw and Pi seem to remove that from the equation, letting you keep your 'history' and customization while using whatever inference provider you want. If Muse takes off, which seems to be built upon or atleast arch'd similar to OpenClaw, I think we'll see the rise of on-device harnesses.
This would further the efforts to "resist all attempts to.... lock them into your service, and will drop you if your competitor makes a 3% better product, and will reject every upsell for actually profitable services.", imo. In a model-agnostic harness all you care about is speed, accuracy, and price.
> If Muse takes off, which seems to be built upon or atleast arch'd similar to OpenClaw, I think we'll see the rise of on-device harnesses.
> This would further the efforts to "resist all attempts to.... lock them into your service
Should make you "happy" to see that Nvidia's new safety feature-set might prevent you from running open models in the newr future, then...
https://nvidianews.nvidia.com/news/open-agent-safety-platfor...
> habit of being aware of nefarious practices
Not all Nerds are the same. I've been attacked for questioning why some companies still use Oracle, when most of the ones I've worked at either migrated off Oracle or were in the process of doing so.
A lot of people in tech buy every bit into cargo culting and following what they've heard is popular or "what everyone does", without much justification or understanding.
Case in point: all the people who still loudly say everyone should jump from Oracle owned MySQL to MariaDB, in spite of MariaDB Inc doing practically everything the original MariaDB fork was meant to "protect" users from at the hands of Oracle.
I'm not saying oracle isn't a huge faceless corporation that wants nothing but money. I'm saying that tech people are IME better at following trends than doing real research themselves.
3 replies →
Everyone gets 'got' once
> They also have a nasty habit of....
This is extremely not the case.
> They also have a nasty habit of being aware of nefarious practices, will resist all attempts to and harvest their data
Lol what? Most of the hackernews gang consider themselves nerds and they lap this shit up! Every new overpriced and unnecessary product that gets released by google, meta, openai, anthropic, whoever, they lap it up!
The usage of Chrome and even worse nitpicking of Firefox is such a strange thing about this community.
1 reply →
I think there's a difference between tech enthusiasts and nerds. Sometimes they might overlap a bit as the technology and the commercialisation overlaps (e.g crypto Vs nfts) but I don't think they're the same thing. My napkin theory is that anything that makes you the end user in their product roadmap (e.g Claude, Ubuntu, AWS) gets tech fans while anything that might not be a fully coherent product but has you figuring out using it gets nerd fans (e.g open models, Debian, Home labs)
Not really. I use cloud AI a bit for development but I don't put personal data in it. That only runs on my local systems.
I also swap all the time for whoever is cheapest
People jumped ship from OpenAI because of their involvement with the US Government / military / Department of War / what have you. That spike was enough that Anthropic was starting to noticeably struggle, which is why they had to rent compute from X AI to get their compute back up to normal, I honestly think if they didn't hit this wall they might have IPO'd much sooner, before they bought extra compute from X AI they downscaled how much compute you can use, but it was poorly done because people got used to much higher limits, they should have explored more strategic options, their changes also broke my workflow several times over. As a result of anthropic trying to deal with the bleeding some users left for OpenAI because it was "unlimited" for some time, then they added limits too.
> AI companies are bad at making software
They should try using agents. I hear they can write great software.
I know that's a joke, but: their software sucks because it's heavily vibecoded
Claude Tag could actually be really useful. Unfortunately, it's too much of a black hole for money. I tried adding it to incident channels, but if a channel gets left open for a few days, Claude will find ways to burn tokens waking up with empty prompt caches and doing nothing. $400 burned by Sonnet 5 on a single incident created because some alarms were oversensitive and didn't distinguish faults from errors. And I still had to prompt it like 5 times to get it to adjust metrics and tweak alarms correctly. Absolutely insane for a change that I could've made as a human in 10 minutes or had a directed Claude session under me do it in 2.
Did Claude actually lose the lead? They definitely lost a lot of good will but the only people I hear talking about actually switching away are people on message boards. Hermes/Openclaw users did as well but it was always reluctantly to something worse. At work it's still very much Claude first and only sometimes others if Claude fails, which is increasingly less often
The people who switch back and forth between providers every month are a very small, but loud, minority.
OpenAI did pull ahead in limits and quality for a while. Anthropic took it back with the Opus 5.5 rollout. I maintain subscriptions to both providers and use both daily. I can confirm these differences were real, not "it's just vibes and nobody knows anything".
I would bet that 99% of each company's paying customers either did not notice, or did not care enough to consider changing.
The concern for them is if that small group expands. Moving back and forth limits perceived lock in. Someone who moves between oai and Anthropic will also move to e.g. an open source when it’s good enough for their purposes, or will start experimenting and combining. This is how the market will slip through the attempted control.
This is me and after using Muse which is free, I have not been hit with any usage limit messages and was able to build a simple iPhone app with it. It makes me question why I am again paying $20 a month for chatGPT. I did for a long while then canceled but went back to paying just a few months ago.
> the only people I hear talking about actually switching away
Opus 5 had really bad writing so I switched to OpenAI, though Opus 5.5 largely addresses that and it's not like them building some Slack integrations (or any other non-core stuff) halts the actual model training in any capacity. For what it's worth, Astra is a pretty good model and for all I know the new Sol will be as well, it's just that it's getting more expensive.
They've definitely lost many experienced devs who can see Anthropic's offerings are insuperior to OpenAI/Codex; I know I am not the only one.
insuperior?
2 replies →
I think it's pretty subjective if you mean "lead" to be capability and not raw number of users. I jumped back and forth quite a bit last year because there were some pretty major shortcoming in both. Now they're both quite reliable without too much hand-holding. Codex now consistently works better for the kind of work I'm doing, and it's good enough that I'm not inclined to go to claude, because I don't have any significant problems.
> unnecessary products no one asked for (see Claude in Slack),
Hey, I never asked for it, but Slack Claude (ie. Claude Tag) has actually turned out to be a useful tool for a few things.
I agree. We use that a lot and is working out pretty great for us.
We use it for quick research that other teammates can follow, filing bugs, quick first round investigations on incidents, etc.
What uses has it presented? (Claude in Slack)
Basically an easier way of doing things we could probably do with custom scripts.
For instance, we rotate reviews (we post to Slack to ask for code reviews) between reviewers on a team with Claude Tag, and we also remind reviewers (in Slack) if it's been more than 24 hours since the review went up and they haven't started it yet.
"AI companies are bad at making software"
Claude Code and Claude Design would like to have a word. Absolute killer products.
Claude code is a terribile harness, visibility is below zero, require a terminal session per each workspace, remote access makes you wonder if you should give copilot deluxe + a try. And according to the TOS you can't you whatever you want
> And according to the TOS you can't you whatever you want
Isn't that every major AI company?
> AI companies are bad at making software
Isnt Chatgpt one of the most successful product of all time?
> Isnt Chatgpt one of the most successful product of all time?
Lets assume that is true[1], all that says is that OpenAI has the best marketing of all time.
The best products are frequently not the highest-selling.
-------------------------------
[1] Depends on how you are measuring "success". If you're measuring it by revenue as a percentage of all products, it's probably not even in the top-ten.
Depends on what your metric is. If you care about profitability it could turn out to be the least successful product of all time.
something selling a lot doesn't mean it's good. Is McDonald's good food? by some metrics of course it is. But if someone says McDonald's makes bad food, I know what they mean
By software they mean "application". Obviously chatgpt is software.
Chatgpt is an application for the underlying model. Most users dont care about the model they care about Chatgpt the App.
Codex is for a small software/programming market.
To survive they need to capture different wide population markets. Can't really fault that logic. The whole point of "SI", is general purpose right? So that would imply being used by multiple markets with multiple products.
> OpenAI won a lot of good favor for the generous Codex subscription and the efficiency of their models, but now that many people have switched over from Claude, they think they can leverage their position to peddle a stream of unnecessary products, and crack down on the generous limits[1] that brought everyone to Codex in the first place.
They're both finding out, very painfully, that there is no moat.
They attract customers by selling at a loss, but that only works when you can turn the dial up on those customers and start selling at a profit.
If either of them had a moat, this would work. Neither of them have a moat.
We should be grateful there are at least two serious competitors, and hope for more. (Come on Europe/Mistral, please do something interesting . . . )
If ever the competition reduced it would turn into an absolute shitfest of nonsense very very fast, and that is a prime reason not to allow them to "pace the frontier".
I have been waiting for the next generation of mistral models for so long now. They were supposed to come out this summer… I hope the delay is because they’re making something better, not just because they’ve fallen too far behind.
Mistral models were the best for my use case when they came out, but they’re mostly almost a year old now. Hard to keep justifying using them, especially with all the price cuts this year on other models.
I was hoping India would be in the race too but all they have is Sarvam that seems like a model from a year ago.
> Come on Europe/Mistral, please do something interesting . . .
lol. At this point, they're miles behind home-appliance-manufacturer Xiaomi.
(Admittedly Mimo v2.6 is legitimately quite good, and really pushing the frontier in certain respects. For e.g., it's the only music generation model that actually listens to instructions.)
Don't know why this is grayed out... europe does not matter for AI. Nobody thinks or cares about them. Mistral makes some good OCR models but that's it. Nobody is making products primarily for the euro market and euros make nothing even equivalent to months old cheap chinese models. Europe is completely irrelevant.
The generous subscriptions are wildly unprofitable and both companies are just throwing stuff at the wall to see what sticks, it's that simple.
The "generous subscription" was actually just heavily subsidized usage.
People seemingly ignore how ruinously unprofitable those companies are.
The interesting part is this:
"After losing a bunch of customers to Codex"
Hot take: Right now it's quite simple to loose customers, like customer churn from A to B or back. That's the weakest spot, isn't it? No matter how complex the products grown and how deep they are integrated into our systems, common users can easily switch. Even heavy users, I argue. Just tell Codex to rewrite the existing Claude-instructions into their own. Even that may not be necessary, if you organized your work "agent agnostic".
And I dont see how that could change, that's why they try to offer tools that are even more integrated into our lives. Like "dots". But this also narrows down the use cases and client base, I argue.
They're just trying things. It's not a conspiracy to "distract" anyone.
I don't think it's a conspiracy, they're just making the same mistake many other software companies have in the past. When you sell your products to the most discerning and well-informed consumers, you enter into a cutthroat race to the bottom. OpenAI would like to diversify to more profitable ventures, but since ChatGPT they have yet to release something novel that was truly successful, nonetheless profitable, and thusfar nearly every experiment has been a flop, so to speak (ChatGPT Atlas, the Sora App, Instant Checkout, etc).
The bitter lesson
The Bitter Lesson is significantly more interesting than watching researcher companies cosplay at ...whatever you call what they're claiming to be doing.
https://en.wikipedia.org/wiki/Bitter_lesson
This is the standard VC enshittification playbook, people predicted this years ago. You subsidize prices with funding until you establish a monopoly, then you raise them as high as your customers can afford. It’ll continue to get worse from here.