Not him but 3 models dominate: GLM 5.3, Qwen 3.8, Mimo 2.6. All censoring certain things. Numbers and other uses are perfectly fine. They are like 0.10-0.15 per 1M tokens. American AI lost the game already, people just can't see it.
> American AI lost the game already, people just can't see it.
Microsoft has been releasing dog shit insanely overpriced software with decent alternatives for decades and is still used in every single company I work for or with.
Your take is the "current year is the year of the linux desktop" meme of "ai"
You're a decent sized company and wants to manage the SW + security on all your employee's PCs. They need to be able to update/remote SW on your machine remotely, see your settings, etc.
I don't think anything comes close to Microsoft's offerings. Macs suck. Ditto Linux.
It could be a Linux vs Mac/Win situation though. Sure the HN crowd will build their own harnesses and hook up cheaper models but the average Joe Sixpack wants something that "just works" out of box and the frontier labs seems well positioned to give that to them.
The difference is that Microsoft did that while relying on the extremely load bearing windows ecosystem. These AI companies have no equivalent lock in, nothing even close to it honestly.
I can guarantee you 80% of people will call any llm "a chatgpt", most have never heard of claude, even less of opus, or sonnet, "deepseek" probably reminds them of a brand of toothpaste or something like that, "GLM" might make them think of the new mercedes SUV perhaps. 99.9% will never self host, nor send a single sent to a chinese model provider.
I don't know a single company using deepseek internally in any capacity, and I have friends in a lot of tech/tech heavy companies, virtually all of them use claude, the lucky ones get cursor with claude/chatgpt/grok.
Once again, just like your other reply, you are mixing things up- a logical fallacy called the straw man; look it up. Your responses usually use this type of misdirection. Coding agent =/= API.
AI does not mean coders coding with agents all day long. AI integration is mostly for data processing, which is the promise and use of the APIs.
Most people I know working in banks etc are using Co-pilot. Not because it is any good, because it integrates with the entire MS ecosystem - management pushes it. I would suspect that is the main driver of MS's AI penetration numbers for desktop-level non-API usage.
When Microsoft did that there wasn't really any competition and they were also the cheap option. I don't think they would survive if there was a sufficiently good chinese OS at the time.
People use them, if for no other reason, because they are cheap, or are part of the Chromebook generation and have gotten used to it
Of their suite, Presentation and Sheets are the only ones people really have gripes about, Sheets by power users because it isn't Excel and it can never be, and Presentations because it's the ugly duckling of the suite
Huh. I’ve never worked at a company that doesn’t primarily/exclusively use Google Docs/Sheets/Slides. Guess that goes to show we’re really all in our own little bubble.
OpenAI and Anthropic have both transitioned into product companies. ChatGPT (the app) and Claude are both one-click installs that just work. People and businesses with pay for this.
People will also pay for the best (or the perception of being the best). Since it's hard to tell what "intelligence" really means model to model, there's a sense of safety in giving a task to the "best".
They are just not overcharging. Nvidia's top AI chip Rubin sells in 72-GPU racks for about $3.5–7.8M. A rack running Xiaomi's MiMo V2.6 Pro generates roughly 150–300B tokens a day, worth about $130–260k at Xiaomi's API price.
That's a payback of the infrastructure in a few weeks in theory. After a few weeks or a month, the only cost is electricity, and whatever they make after that is pure profit. This is why they can charge normal prices. Not $50 for 1M tokens.
The reason why that happens is not really Europe's fault. Any remotely-competent developer just works for US companies, why would they not? Especially for AI, there just isn't enough investments in the EU to offer those salaries.
What's left are incompetent developers paired with people who get government contracts and leech off them. Ask me how I know.
The usage numbers tell a different story. Not only is the great majority of AI users on American frontier providers, they are also willing to pay the premium price for the premium product. It's not like tech is oblivious to the Chinese models. It's that the industry is aware they are always 6 months to a year behind.
You didn't discover some new trick for cost performance. And the rest of the world isn't dumb.
You're just too broke to afford the supercar and justifying the hooptie. It gets you to the kindergarten class after all. And that's all you need.
The problem with being always online is not “not-enjoying” but stopping to think. People are getting used to be feed what they need to think constantly (from politics to product reviews) and then their critical thinking skills atrophy.
Moreover, I’m not sure if being on social media is something “enjoyable” anymore, and it’s more like smoking tobacco where it becomes necessary but is not enjoyed.
She's just arguing that most of the purposes of code reviews should be done before the typing part of coding instead of after. Identifying the best design before investing in typing / tokens.
Waterfall has to do with the size of the loop and when the customer gets to provide consumable feedback. She's not proposing changes to the size of that loop.
I mean, we certainly are in a lot of cases. Spec driven development is waterfall, and we're all in the midst of a new experiment to see if it works. I'm not sold, but consultants love it.
The power of contacts, and looking at the initial list it seems it’s a bunch of founders whose startups succeed using Ruby on Rails, so I guess yeah… they owe some to DHH
Most likely also technofascists and other oddballs helping out their technofascist buddies. One of the other big sponsors:
In July 2026, Lütke expressed support for setting voting rights proportional to the level of income tax paid by a citizen, with those paying no income tax being stripped of their voting rights and others gaining one additional vote for each $100,000 paid in income tax up to five votes per person.
It raises the question: what is a distro since many distros are customizations of other distros. What makes Ubuntu a distro and not just a customization of Debian?
The video above is in inline with my opinion. If you ship an ISO (so users can install your distro directly without installing another distro first) and have control over the packages then you are a distro. Apparently, Omarchy has its own ISO and it's own mirror of the arch package repo so it is a distro by that definition.
Steamos is a mirrored snapshot of arch (like omarchy) + bash and python scripts (like omarchy) + their own shell (steam running on top of gamescope for steamos and a custom quickshell based one running on top of hyprland for omarchy) + modified upstream kernel (both steamos and omarchy)
At the time Ubuntu started, Debian's old installer (boot-floppies) was legendary for its unfriendliness, and the new all-singing, all-dancing replacement was a mammoth endeavour and a long way off completion. Ubuntu provided a simple, friendly installer that covered only the limited set of the most common use cases, but was a dramatic improvement for new users.
They also employed a bunch of Debian developers, and gave them space to continue their work for Debian. It wasn't all smooth sailing, but Debian was stronger as a result of their contributions right from the very start.
Now, maybe Omarchy are providing a similarly major contribution to Arch - but, if they are, they've kept quiet about it so far.
No, Israel has maintained relations with the Kremlin and has refused to impose sanctions against them, as well as rejected calls to send military aid to Ukraine.
Using local models, usually means generating content on a more private, local device with the weights of the model stored on that device - a local LLM model could have helped write these posts. It only gets easier to self-host increasingly capable models for generation of text, images, video etc. E.g: Prism ML Bonsai have an older 27B text generating LLM model squashed down to a few GB at what they describe as 95% quality, which can run on a phone.
reply