dbgrman

That it has a name and a number, but it doesn't say what the differentiation is.

Gemma = small model Gemini = big model

Gemini 4 = latest model.

Gemini 4 Aragon = ?? Gemini 4 Carbon = ??

> Building on Gemini 4 Argon, which was launched two weeks ago, Carbon illustrates the pace at which we’re pushing the frontier forward. Our researchers and models are working hand in hand to advance capabilities faster, while strengthening our safeguards in tandem.

Is it a variant of Argon? A different model? The next version? It seems like this is a separate class (like Anthropic's Mythos).

anyway, its needlessly confusing.

If its just tedious, I bet there is room for agentic/automation to keep things tidy.

alfred and quicksilver etc. all work surprisingly well.

Let's assume for a second that the intelligence is there and these models really are great. Would it make sense to hire CTO of some big-tech company to do code reviews for your startup? Feels kind of like overkill to me. Code review is not about more intelligence. To me, it's about more cultural context. And all things equal, what difference would the model make, at anything above sonnet5 medium level?

"That's why all the work we do in growth is justified. All the questionable contact importing practices. All the subtle language that helps people stay searchable by friends. All of the work we do bring more communication in. The work we will likely have to do in China some day. All of it."

BOZ. Meta's current CTO. Not too long ago.

lol no thanks. Even with a proven "superintelligence" A.I, I'm not giving Meta access to anything. And that says something coming from an ex-Meta engineer (Ig/whatsapp/fb).

Another option that works quite well is called FAFO. So, I'd say just publish it, and we'll see. Keep us posted!

from meta's own statement: "The majority of the terms are required to remain in place for 10 years, but our Time Limit and Night Mode features will start with a five-year commitment. However, if industry peers sign on to the agreement, it will both extend this commitment to 10 years and prompt stronger default limits — reducing the Daily Limit to one hour per app and expanding Night Mode hours to 10:00 PM–7:00 AM (up from midnight–6:00 AM)."

I'm at loss for words at how shitty this paragraph reads. If anyone has any hope that Meta (and the rest) are even slightest concerned about the problem at hand (mental health of children), then they're very dim.

Meta literally goals on creating high-quality teen producers and consumers. Every single experiment, every single project is optimizing for that.

At least now its clear that behind the fancy marketing, its all "just business" for them and if they could, they'd sell the kids' soul to create "more shareholder value"

Hmm... interesting. previously this was annas-archive.pk, now its on gl domain. What happened?

I don't think most of it has anything to do with intelligence and what I need the AI to do. In our own Claude.md files, we aggressively cut down on things that are irrelevant. Claude code has rules/ feature. Most of the content of Opus 5's system prompt can probably live in a rule or a progressively disclosed file (e.g. when asked about Mythos, read [md file link].).

In my personal experience, the claude.md file heavily influences the LLM's responses. It makes judgement from any and all information you give it, and try to tailor the response to fit the stereotype it gave you. After deleting my claude.md file (ironically, on the advise of Boris), Claude Opus 5's language slop issues drastically went down.

> Claude avoids saying "genuinely", "honestly", or "straightforward". Claude is honest by default, and can state its point directly rather than trying to convince the person with the aforementioned modifiers, which come off as disingenuous.

And honestly, that's the most useless thing in Claude.md file because literally every paragraph has an "honestly" in it!

* No one, not even E. B. White wrote the final document in a single pass. With dynamic workflows, you can now implement a writer's workflow.

* Opus pays more attention. So anything in your Claude.md, your code's claude.md, in Claude Desktop, the customizations, even your name, will be used as context. If Claude knows you are a mechanical engineer and trying to write code, it will try to write code and explain it to you in some stereotypical way you did not expect.

* There are problems that require horizontal scaling and not vertical, even in intelligence. If I want to serve tea to 200 people at my home, I just need 10 decent adults, not Gordon Ramsey. So if your problems demand horizontal scaling, a dynamic workflow with Sonnet 5 medium with 200K context window will be more productive than Opus 5 max at 1M token context window.

Kudos to France. But the problem isn't telemarketing calls. It's the wholesale grab-fest of phone numbers from every.damn.nook.and.crany of the US soil. Hospitals, Car dealers, dentists, vets, brokerages, title offices... literally EVERY one of them is leaking YOUR information to everyone in the world.

within a week of closing a deal on my house i get a scam paper mail that almost looked legit asking me to pay $100 to get the title documents for my house. There aren't that many parties involved in the sale.. and its one of them (or all of them) broke the trust and bartered my information.

nearly all of them are uploading their contact list to meta and google.

THIS practice has to stop.