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

Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

tokens.billchambers.me

Post reply on HN
Print view View original

261–270 of 620 posts

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#261

Post by aerhardt »

colechristensen wrote: post #250
aerhardt wrote: post #235

Earlier quoted context omitted.

> What's your bar that needs to be passed I provide four examples in my comment...

Your qualification for if an LLM can write a novella is it has to be as good as The Metamorphosis ? Yes, those are examples of novellas, surely you believe an LLM could write a bad novella? I'm not sure what your point is. Either you think it can't string the words together in that length or your standard is it can't write a foundational piece of literature that stays relevant for generations... I'm not sure which.

I don't think it can write something that's of a fraction of the quality of Kafka.

But GP's argument ("limit the space to text") could be taken to imply - and it seems to be a common implication these days - that LLMs have mastered the text medium, or that they will very soon.

> it can't write a foundational piece of literature

Why not, if this a pure textual medium, the corpus includes all the great stories ever written, and possibly many writing workshops and great literature courses?

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#262

Post by dirasieb »

sergiotapia wrote: post #256
dakiol wrote: post #46

We dropped Claude. It's pretty clear this is a race to the bottom, and we don't want a hard dependency on another multi-billion dollar company just to write software We'll be keeping an eye on open models (of which we already make good use of). I think that's the way forward. Actually it would be great if everybody would put more focus on open models, perhaps we can come up with something like the "linux/postgres/git…

I can recommend this stack. It works well with the existing Claude skills I had in my code repos: 1. Opencode 2. Fireworks AI: GLM 5.1 And it is SIGNIFICANTLY cheaper than Claude. I'm waiting eagerly for something new from Deepseek. They are going to really show us magic.

it is also significantly less capable than claude

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#263

Post by foobarchu »

oremj wrote: post #205
ljm wrote: post #118

Earlier quoted context omitted.

Not so much atrophy as apathy. I've worked with people who will look at code they don't understand, say "llm says this", and express zero intention of learning something. Might even push back. Be proud of their ignorance. It's like, why even review that PR in the first place if you don't even know what you're working with?

Is this the same subset of people who copy/paste code directly from stack overflow without understanding ? I’m not sure this is a new problem.

It's a new problem in the sense that now executive management at many (if not most) software companies is pushing for all employees to work this way as much as possible. Those same people probably don't know what stack overflow even is.

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#264

Post by tossandthrow »

dgellow wrote: post #156
tossandthrow wrote: post #113

Earlier quoted context omitted.

You are aware that using eg. Github copilot is not one shot? It will start an agentic loop.

Unnecessary nitpicking

Why?

One shoting has a very specific meaning, and agentic workflows are not it?

What is the implied meaning I should understand from them using one shot?

They might refer to the lack of humans in the loop.

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#265

Post by foobarchu »

trinsic2 wrote: post #230
ori_b wrote: post #123

Earlier quoted context omitted.

I suggest you read the sales pitches that these products have been making. Again, when I say that this is the selling point, I mean it: This is why management is buying them.

Sales pitches dont mean jack, WTF are you talking about?

Sales pitches are literally the same thing as "the selling point".

Neither of those is necessarily a synonym for why you personally use them

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#267

Post by danny_codes »

andai wrote: post #101
smt88 wrote: post #86

Earlier quoted context omitted.

Tell that to oil and defense companies. If tech companies convince Congress that AI is an existential issue (in defense or even just productivity), then these companies will get subsidies forever.

Yeah, USA winning on AI is a national security issue. The bubble is unpoppable. And shafting your customers too hard is bad for business, so I expect only moderate shafting. (Kind of surprised at what I've been seeing lately.)

It’s considered national security concern by this administration. Will the next be a clown show like this one? Unclear

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#268

Post by ojosilva »

esafak wrote: post #172
aliljet wrote: post #114

Earlier quoted context omitted.

What open models are truly competing with both Claude Code and Opus 4.7 (xhigh) at this stage?

GLM 5.1 competes with Sonnet. I'm not confident about Opus, though they claim it matches that too.

I have it as failover to Opus 4.6 in a Claude proxy internally. People don't notice a thing when it triggers, maybe a failed tool call here and there (harness remains CC not OC) or a context window that has gone over 200k tokens or an image attachment that GLM does not handle, otherwise hunky-dory all the way. I would also use it as permanent replacement for haiku at this proxy to lower Claude costs but have not tried it yet. Opus 4.7 has shaken our setup badly and we might look into moving to Codex 100% (GLM could remain useful there too).

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#269

Post by ACCount37 »

gsleblanc wrote: post #142

It's increasingly looking naive to assume scaling LLMs is all you need to get to full white-collar worker replacement. The attention mechanism / hopfield network is fundamentally modeling only a small subset of the full human brain, and all the increasing sustained hype around bolted-on solutions for "agentic memory" is, in my opinion, glaring evidence that these SOTA transformers alone aren't sufficient even when yo…

You probably are.

The "small subset" argument is profoundly unconvincing, and inconsistent with both neurobiology of the human brain and the actual performance of LLMs.

The transformer architecture is incredibly universal and highly expressive. Transformers power LLMs, video generator models, audio generator models, SLAM models, entire VLAs and more. It not a 1:1 copy of human brain, but that doesn't mean that it's incapable of reaching functional equivalence. Human brain isn't the only way to implement general intelligence - just the one that was the easiest for evolution to put together out of what it had.

LeCun's arguments about "LLMs can't do X" keep being proven wrong empirically. Even on ARC-AGI-3, which is a benchmark specifically designed to be adversarial to LLMs and target the weakest capabilities of off the shelf LLMs, there is no AI class that beats LLMs.

Re: Anonymous request-token comparisons from Opus 4.6 and Opus 4.7

#270

Post by danny_codes »

coldtea wrote: post #10

This, the push towards per-token API charging, and the rest are just a sign of things to come when they finally establish a moat and full monoply/duopoly, which is also what all the specialized tools like Designer and integrations are about. It's going to be a very expensive game, and the masses will be left with subpar local versions. It would be like if we reversed the democratization of compilers and coding toolin…

I doubt that’s the case. My guess is we’ll hit asymptomatic returns from transformers, but price-to-train will fall at moore’s law.

So over time older models will be less valuable, but new models will only be slightly better. Frontier players, therefore, are in a losing business. They need to charge high margins to recoup their high training costs. But latecomers can simply train for a fraction of the cost.

Since performance is asymptomatic, eventually the first-mover advantage is entirely negligible and LLMs become simple commodity.

The only moat I can see is data, but distillation proves that this is easy to subvert.

There will probably be a window though where insiders get very wealthy by offloading onto retail investors, who will be left with the bag.

Post reply on HN