>wait… if you’re burning through your Codex limits, why not just use Dots? Because I'm in the EU

Bro with the hard questions. I don't even know what I did the last few years.

depends what you are doing... for scientific shit i agree 100% but if you want to let your anime waifu's dance... oh boy than it is not even close on a serious note... video editing is on another level opus piped.video/watch?v=3PY9Zk8p… vs astra piped.video/watch?v=724Pehux…

The benchmark saturates if the solver manages to create a processor that lets the solver run faster than the instance it currently is running on. Then the universe terminates because division through 0 is just child's play in contrast.

$$$ or % of weekly on your plan?

how about HWE Bench hwebench.com/ ? In theory it'll never saturate since you can always improve processor design

Old man yells at cloud. Embrace the looping future.

If there is actual proof of it, sure. But so far, I've never seen any, except for the AMD vs. Anthropic drama. Reddit and X delusions are not proof. We benchmark harnesses and endpoints basically daily, we have also not seen proof of it.

The train never stops. Choo choo. Next stop: singularity. The last time we had a moral panic, we didn't have a player like China that doesn't share our doomsday delusions and is technological almost on par. Keeps us aligned. :)

How tf is Dario not constantly dooming lol - cue in point "white collar blood bath" and the ipo prospectus that is busy explaining over 80p how dangerous the shit is you should pls buy stocks for.

Made with Claude Opus 5.5. The fearmongering about AI always makes us forget that NOTHING WENT FOOM, as was always predicted. Send this to your doomer friend who has a very high P(Doom). Accelerate.

You mean like he tried to formalize how GPT-4o was the absolute limit of what LLMs can reach?

On a serious note, at the beginning of all this, I really envisioned something like an adapter marketplace. Imagine being Terence Tao, working with ChatGPT for years and gradually putting so much of your own knowledge and data into it that you can press a button in the UI to build a "Tao LoRA" with incredible math expertise. Then other people could pay a few bucks to use that "Tao LoRA" through the marketplace. It's basically shareable, semi-online fine-tuning with a business model attached, and I still think you could absolutely make that work. But instead we got GPTs... lol.

Excuse me, m'lady, but we connoisseurs of generating anime waifus discovered the power of LoRAs and adapter modules ages ago. Once again, waifu technology remains years ahead of enterprise AI.

Haha, ganz gewiss.

a big chart with company logos obv

I don't want to imagine Jensen gooning at anyone ;_;

I don't read/write code these days, I only instruct an LRM. But it's not because I think LRM code quality is perfect, or even good. And it's not because I think my instructions are always being perfectly acted upon -- I know they aren't. The thing is: LRMs don't just write your code, they unlock new ways to control what you're doing and maintain understanding of your system, which are about as effective as reading code. You can better test your codebase, you can audit specific components, you can spin up visualizations, you can ask the LRM to red team, etc. You can do all of these things at high speed -- you couldn't before. The benefits of reading/writing code can now be achieved in new ways. All in all, the ROI of writing code by hand is no longer looking good, not because LRMs got perfect but because they're so fast at doing things that you can develop new workflows around them, workflows that end up being more productive than the old workflows.

yes got the same

Imagining the Big Yud in an orgy did more damage to my well-being than AI ever has or ever will.