They’re calling this out because Anthropic is afraid of dirt cheap, “good enough” open weights models undercutting them. Probably very afraid now that even Nvidia is on that boat, with huge Nemotron models.
The real battle isn’t pro AI vs anti AI. It’s closed weights answers-as-a-premium-service vs open weights, hackable tools. It’s Huggingface vs OpenAI. It’s akin to Lemmy vs Reddit.
Why would anyone use Anthropic once people figure out LLMs are configurable tools, not “AGI,” and efficient ones cost like 2 orders of magnitude less to run?
So they want to squash open research. Because businesses are asking about costs now, they don’t realize they can just host assistants on-prem or through dirt cheap competing providers, but they’re starting to figure it out.
Deepseek is going to eat their lunch. They are not that far behind and cost pennies in comparison.
Not just them. GLM, Qwen, Kimi, Stepfun, Baidu’s models. Z-Image. Small finetuners, Huawei’s prototype. There’s even a Chinese fast food chain that trains a ridiculously good audio/text mixed model (Longcat).
I actually thought the recent Deepseek preview was a little underwhelming and “deep fried” compared to competition, though maybe it’s just underbaked. And the architecture is interesting.
Gemma is great, too, if Google would actually unrestrain it and give it Gemini’s architecture.
Europe is struggling though. Mistral (and everyone else) basically can’t do anything because the EU left regulation ambiguous; however strictly they regulate AI (and it should be pretty strict), anything is better than “we have no idea if we’ll get litigated, the law is clear as mud and might change?” They have at least one communal training project too, but everything I’ve seen is weirdly dated, architecture wise, like they’re living two years in the past.
“We only want things to go fast when we’re in the lead.”
Fuck these poseurs.
That sort of implies they have been dethroned in some way, but the opposite seems to be true. They are eating OpenAI’s lunch. Claude is widely considered the best development AI, and they just passed OpenAI in business customers.
That said, they do post the most ridiculous blogs at times. Actually, I have yet to see a blog post by them that wasn’t complete bullshit, but that’s probably just selection bias. I assume they have some employees who aren’t complete morons.
Every statement this shitty company releases is exponentially more cringe. I’m calling for a slowdown in Anthropic press releases.
What did they expect? They fucking bought all the hardware in the world, of course it won’t go faster now. Fuckers!
This is the hype train, claiming they can and will do so much more than be junior software engineers.
Why would they want a slowdown while simultaneously getting ready for an IPO?
Because for one, these things can’t and won’t do so much more than be junior software engineers because a complete lack of understanding of how the human mind does not inhabit the body or pilot the body, but the mind separated from the body loses wholeness. That’s going to blow up these IPOs if they can’t actually deliver, so the slowdown is the industry all shaking hands and saying “we need more time for this charade to hold up” when actually none of them will slow down, they’ll just be trying to hide how little forward progress there is to be made left with these kind of models and without moving to something even more complex. They need the con to last as long as possible in hopes of delivering something useful at all to justify the massive increase in surveillance capitalism that citizens obviously do not want. Their IPO will crash if they get found out to be frauds right after going public, and they aren’t satisfied enough with a quick pump and dump, they all want to be the next Musk. So they need a global slowdown to justify how things have actually hit a wall in terms of making LLMs useful to every day people.
Anthropic’s opus 4.8 is well beyond a junior dev. Most of the time it outperforms senior devs. Then there’s mythos we don’t have access to yet. We’re fucked. Even if human R&D stops now
It’s both smarter in many ways, and extremely dumb in others, even in a really good and secure tool harness (which is rare).
It’s a great assistant and task finisher, but the architecture is just fundamentally not suitable for handing any significant responsibility.
IBM 1979 training manual
I disagree. We should replace the C Suite Immediately. Lazy, entitled and unaccountable and they are paid WAY beyond their value.
Even worse: nested API calls to ever-changing black boxes with literally randomized outputs.
