Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GPT-5.6 Sol Ultra will be in Codex

twitter.com

Post reply on HN
Print view View original

321–330 of 433 posts

Re: GPT-5.6 Sol Ultra will be in Codex

#321

Post by Supermancho »

alfiedotwtf wrote: post #233
bob1029 wrote: post #190

Earlier quoted context omitted.

I'm struggling with the assertion that these models cannot provide reasonably deterministic guarantees. I am using gpt to populate JSON objects conforming to a list of natural language constraints for purposes of generating fake customers. I am finding that gpt5+ never fucks up. Not even a little bit. I've ran this test hundreds of times with 20+ constraints and it's been perfect every time. Stable information yields…

> I'm struggling with the assertion that these models cannot provide reasonably deterministic guarantees. LLMs are probabilistic by design so running the same prompt multiple times will give you different results. Otherwise, we wouldn’t needed LLMs and could replace it all with Postgres

> LLMs are probabilistic by design so running the same prompt multiple times will give you different results.

Reasonably deterministic is the phrase. If I can be sure the LLM is giving me back the same result 99% of the time I need it, that's reasonable for me. Maybe this is not reasonable for others.

ie How often will an LLM get 2+2 wrong? Now expand until you're uncomfortable.

Re: GPT-5.6 Sol Ultra will be in Codex

#323

Post by LollipopYakuza »

cactusplant7374 wrote: post #315
Szpadel wrote: post #199

there seems to be very big misunderstanding about what the "ultra" is, so let me explain it basing on the codex source code: it's similar to Claude code ultracode. there is no ultra effort level implemented on the backend. it's just alias in the codex to max effort setting and single line addition to prompt to use subagents proactively. that's all as far as we know pro models work differently. for once those are back…

That's strange. One can easily steer their session to use agents proactively.

Many features of the cli tooling of these providers can be achieved by prompting.

The way I see it is that they try to normalize and ease the use of practices established by the community.

Re: GPT-5.6 Sol Ultra will be in Codex

#324

Post by killix »

d4rkp4ttern wrote: post #264
andai wrote: post #64

Earlier quoted context omitted.

Wow, I hadn't heard of this! const audits = await pipeline(found.files, file => agent(`Audit ${file} for missing authentication checks.`, { label: file }), ) I asked Claude in the browser if it could do anything like that. It wrote a little frontend app that calls the Anthropic API (with fetch()), without including a key. I expected that to fail, but it worked! Apparently in the web chat (and also in Claude Code?[0]…

My understanding is that in the generated JS code, the subagents are invoked as headless Claude, equivalent to Claude -p, or the agents SDK.

[flagged]

Re: GPT-5.6 Sol Ultra will be in Codex

#326

Post by tekacs »

behnamoh wrote: post #14

I still don't know why OpenAI doesn't put gpt-5.5-pro in Codex. It's one hell of a model and easily parallels Fable/Mythos. Sure, it'll use up your quota much faster but that's the price some users are willing to pay for absolutely high quality responses. I think gpt-5.5-pro runs 12x parallel gpt-5.5 agents behind the scene and uses OpenAI's secret sauce to synthesize their answers into one insanely good response.

https://developers.openai.com/api/docs/models/gpt-5.5-pro

> GPT-5.5 Pro does not offer a cached input discount.

I think this tells you in one line. It's basically set up for one-shot inference right now, by the looks of things. If you use this in a harness, it would almost immediately fall apart on cost. Not to say that they couldn't make it work, just saying that at least as it's delivered currently, they haven't done so. On the web, there might be doing something to get the equivalent of that behavior internally, such as keeping the session truly alive on GPUs rather than using their external-facing cache-style approach.

Re: GPT-5.6 Sol Ultra will be in Codex

#327

Post by noduerme »

someguyiguess wrote: post #213
noduerme wrote: post #187

Earlier quoted context omitted.

Other than the delay time, I'm not sure I see the difference. You're removing your primary from the job of writing code, putting them into an editorial role, which removes responsibility and agency and actual hands-on understanding. The quality of the code is beside the point. More friction (language barriers, time zone difference) is actually better if you want to maintain institutional knowledge, because it require…

I reject the premise that using LLMs absolutely leads to loss of institutional knowledge. It is trivial for an LLM to generate a knowledge base of any kind in any language which can answer any question about your institution at any time. How is a bunch of fragmented humans with limited knowledge who can’t all communicate with each other better than that?

Have you ever taken the time to read hundreds of pages of documentation to fully understand a massive codebase? Neither have I. You learn it by working with it all day, and you're careful with it. A complete "knowledge base" of business logic is, itself, indistinguishable from code. Code that no one can or will read and learn unless they have to be immersed in it.

So the "knowledge base" an LLM generates is not useful as code, nor is it useful to humans. It may be useful to other LLMs as a lossy compression scheme for the original intention of the business logic and code.

That's not even in the same universe as institutional knowledge. Handing any serious business to something like that is malpractice, and sloppy beyond belief.

I'm not surprised that a lot of people think this way, because they never really grasped the benefit of holistic business knowledge united with code to begin with. Those people always outsourced, always got shitty code, and never really unified their systems. Cheap people and cheap companies take cheap exits. That's fine. But yes, many fragmented humans who all understand their portion in depth is much better than a bunch of markup files at retaining the knowledge of why things are done, procedurally, the way they are.

Once every six months the CEO calls me and asks me to remind him why our software does something like, idk, create a reverse payment instead of voiding a charge in some situation. Or some other thing he has asked a dozen times before. And I know the answer to why, or I know where to look, because I was in those meetings 5 years ago or because I wrote the code myself and asked the question when I wrote it.

Institutional knowledge is a very large context window, if you need to think of it that way, and LLMs are a shitty compression method for that. They can tell you how, but they don't reason well with why.

Re: GPT-5.6 Sol Ultra will be in Codex

#328

Post by Supermancho »

solenoid0937 wrote: post #152
senordevnyc wrote: post #72

Earlier quoted context omitted.

I find this kind of cynicism fascinating tbh. On the one hand, it seems so relatable in some ways, because there is something uncomfortable about being seen as naive, in a way that being seen as cynical or negative doesn't seem to carry. I guess it's just self-protective, almost like some kind of perverse Pascal's wager: it's better to think everyone is horrible and be wrong than to think the opposite and be taken ad…

It's a lot easier to sound smart on the internet if you're a bitter cynic. Lots of nerds for some reason have made cynicism a personality trait. They think optimism/honesty is hopelessly naive, therefor cynicism is the correct default.

> Lots of nerds for some reason have made cynicism a personality trait. They think optimism/honesty is hopelessly naive, therefor cynicism is the correct default.

It is the result of experience. Working with and creating systems (even embarrassingly simple ones), then seeing them fail in a myriad of ways more often than succeeding, colors your expectations about throwing humans into the mix.

Children learn to lie as part of their natural development, but do not always externalize that until faced with media (Airheads candy commercial or equivalent). Either way, honesty is expected as a default for utility and not an expectation in leveraging goals.

Re: GPT-5.6 Sol Ultra will be in Codex

#329

Post by stingraycharles »

alightsoul wrote: post #274
stingraycharles wrote: post #195

Earlier quoted context omitted.

Ok I’m not sure I follow your point here. Isn’t all that he’s saying that if they find some optimization techniques, that gives them an edge? And that makes sense? How is this suddenly evidence of him being a villain?

The evidence is that unlike Deepseek he does not publish his compute multipliers. Under that argument Deepseek should not publish any of their research either.

But Deepseek has a very different way of operating their business as the underdog. They also publish their models as open weight, which Anthropic also doesn’t do.

I don’t think this makes Anthropic a villain?

Re: GPT-5.6 Sol Ultra will be in Codex

#330

Post by stingraycharles »

Der_Einzige wrote: post #300
solenoid0937 wrote: post #221

Earlier quoted context omitted.

It has got to be one of the most insane takes I've read on HN, which to be fair has been trending towards "unhinged" when it comes to Anthropic and AI safety. Compute multipliers are like a quant firm's trading algorithms. They're the crown jewels, the whole alpha of the lab. If you leak them, the lab dies. Protecting them does not make Dario a villain, it's literally his job. It's also Sam's job, Denis's job, Mira's…

The fact that they don’t release model weights for free to download on huggingface means that Sam, Mira, Dario, etc are ontologically evil and may they all reincarnate as either durian fruits or cockroaches, ideally as durian fruit infested with cockroaches…

Is this a serious take or sarcasm?
Post reply on HN