plz1

The data center is still needed, just not the same scale. You don’t run “small” local models to compete with Claude’s size. They don’t publish numbers, but estimates were like 1-2T params for the models like early Opus. You need a lot of hardware to run those efficiently at a scale an enterprise would need.

Or they just spent enough to upgrade developer desktops with RTX Pro’s to run quantized ~100B param models at their desks. Those are very good for coding.

Qwen 3.8 27B, running on a RTX 5090. Works so well.

I didn’t realize the business plans were token metered differently from the Pro plan I’ve been on for a while. $20/month, and it has session/weekly limits, but I rarely hit them, and I use it, a lot.

I do wonder if this will encourage businesses to branch out into running local models for simpler tasks.

Last I checked, submitting false testimony was a crime. If these lawyer stop getting slaps on the wrist, the slop legal work will stop.

Yeah, it’s negligence. They gambled on any paltry fine being worth it, and still being allowed o proceed.

This article is a year old, but relevant because Discord started doing the verifications again. The plan was never canceled, just delayed to a later news cycle. And this will absolutely happen again because there are no consequences.

It’s not RAM slots, it’s unified memory, built into the CPU/chip. ALso means you can’t expand it later.

You’re not getting a 4080 for $200 though…

I mean, Micron for sure did, when they “exited consumer RAM to focus on data centers” at a point where overall RAM was at peak demand. That’s a supreme intent to create artificial scarcity.

So now LEO’s will just buy the data, instead of getting warrants, right?

Hold my beer

-Copilot

The fact that it can do it in the first place means it can do it at all. Just because they require the selection now, doesn’t mean they aren’t harvesting telemetry 24/7, and won’t “accidentally” enable this feature later.

This sounds a lot like fingerprinting under the false flag of making user experience better.

But was the GPU OK? That’s gold right there.

Speech to Text -> NLU -> Ordered summarized -> human for review (we hope?) -> order placed. They are using these to reduce the labor demand on the human portion, and I presume at some point the “to human for review” will just go away, if they get the accuracy dialed in.

The shitty part is, it’s not like they are saving much money, since those jobs pay shit anyways. I have to wonder what the ROI is on replacing the electronics and hooking all that up, plus the R&D that went into the system itself, is.

DeepSeek never said it was permanent in their pricing materials, the article writer did. They are just taking the current expiration date off an existing discount. It’s absolutely a shot across the bow at Claude, OpenAI, et al., but the author was click-baiting, as is tradition.

Wave power is transmitted back to land, usually, it won’t be a floating data center. At least I doubt it would be, but who knows, they are pretty awful people.

caffeinate -d

closes laptop lid

The word you’re looking for is sociapth. You need to be a sociopath to become a billionaire

I mean, it’s shady for headlines, but it’s not false. The board are residents, and the did vote it down. Then they “settled” and I’m betting it was a pittance compared to the data center value. In reality, it was probably a situation where whomever was backing the data center said to throw lawyers and/or money at the problem until it went away.