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

Claude Opus 4.7

anthropic.com

Post reply on HN
Print view View original

401–410 of 1001 posts

Re: Claude Opus 4.7

#401

Post by unshavedyak »

zarzavat wrote: post #336
lanyard-textile wrote: post #200

This comment thread is a good learner for founders; look at how much anguish can be put to bed with just a little honest communication. 1. Oops, we're oversubscribed. 2. Oops, adaptive reasoning landed poorly / we have to do it for capacity reasons. 3. Here's how subscriptions work. Am I really writing this bullet point? As someone with a production application pinned on Opus 4.5, it is extremely difficult to tell ap…

These threads are always full of superstitious nonsense. Had a bad week at the AIs? Someone at Anthropic must have nerfed the model! The roulette wheel isn't rigged, sometimes you're just unlucky. Try another spin, maybe you'll do better. Or just write your own code.

Part of me wonders if there's some subtle behavioral change with it too. Early on we're distrusting of a model and so we're blown away, we were giving it more details to compensate for assumed inability, but the model outperformed our expectations. Weeks later we're more aligned with its capabilities and so we become lazy. The model is very good, why do we have to put in as much work to provide specifics, specs, ACs, etc. So then of course the quality slides because we assumed it's capabilities somehow absolved the need for the same detailed guardrails (spec, ACs, etc) for the LLM.

This scenario obviously does not apply to folks who run their own benches with the same inputs between models. I'm just discussing a possible and unintentional human behavioral bias.

Even if this isn't the root cause, humans are really bad at perceiving reality. Like, really really bad. LLMs are also really difficult to objectively measure. I'm sure the coupling of these two facts play a part, possibly significant, in our perception of LLM quality over time.

Re: Claude Opus 4.7

#402

Post by causal »

Tiberium wrote: post #36
cupofjoakim wrote: post #20

> Opus 4.7 uses an updated tokenizer that improves how the model processes text. The tradeoff is that the same input can map to more tokens—roughly 1.0–1.35× depending on the content type. caveman[0] is becoming more relevant by the day. I already enjoy reading its output more than vanilla so suits me well. [0] https://github.com/JuliusBrussee/caveman/tree/main

I hope people realize that tools like caveman are mostly joke/prank projects - almost the entirety of the context spent is in file reads (for input) and reasoning (in output), you will barely save even 1% with such a tool, and might actually confuse the model more or have it reason for more tokens because it'll have to formulate its respone in the way that satisfies the requirements.

Output tokens are more expensive

Re: Claude Opus 4.7

#404

Post by gruez »

johnmlussier wrote: post #293
gruez wrote: post #286

Earlier quoted context omitted.

>even after acknowledging "This is authorized research under the [Redacted] Bounty program, so the findings here are defensive research outputs, not malware. I'll analyze and draft, not weaponize anything beyond what's needed to prove the bug to [Redacted]. What else would you expect? If you add protections against it being used for hacking, but then that can be bypassed by saying "I promise I'm the good guys™ and I'…

This was Opus saying that after reviewing the [REDACTED] bug bounty program guidelines and having them in context.

Right, but that can be easily spoofed? Moreover if say Microsoft has a bounty program, what's preventing you from getting Opus to discover a bug for the bounty program, but you actually use it for evil?

Re: Claude Opus 4.7

#405

Post by mnicky »

bushido wrote: post #379

I think my results have actually become worse with Opus 4.7. I have a pretty robust setup in place to ensure that Claude, with its degradations, ensures good quality. And even the lobotomized 4.6 from the last few days was doing better than 4.7 is doing right now at xhigh. It's over-engineering. It is producing more code than it needs to. It is trying to be more defensible, but its definition of defensible seems to b…

Maybe this? From the article:

> Opus 4.7 is substantially better at following instructions. Interestingly, this means that prompts written for earlier models can sometimes now produce unexpected results: where previous models interpreted instructions loosely or skipped parts entirely, Opus 4.7 takes the instructions literally. Users should re-tune their prompts and harnesses accordingly.

Re: Claude Opus 4.7

#406

Post by aliljet »

redml wrote: post #245
aliljet wrote: post #64

Have they effectively communicated what a 20x or 10x Claude subscription actually means? And with Claude 4.7 increasing usage by 1.35x does that mean a 20x plan is now really a 13x plan (no token increase on the subscription) or a 27x plan (more tokens given to compensate for more computer cost) relative to Claude Opus 4.6?

a few months ago it was for weekly: pro = 5m tokens, 5x = 41m tokens, 20x = 83m tokens making 5x the best value for the money (8.33x over pro for max 5x). this information may be outdated though, and doesn't apply to the new on peak 5h multipliers. anything that increases usage just burns through that flat token quota faster.

wait. that's insanity. where did you get those numbers from? the 5x plan is obviously the right place to be...

Re: Claude Opus 4.7

#407

Post by cmrdporcupine »

desugun wrote: post #112
buildbot wrote: post #12

Too late, personally after how bad 4.6 was the past week I was pushed to codex, which seems to mostly work at the same level from day to day. Just last night I was trying to get 4.6 to lookup how to do some simple tensor parallel work, and the agent used 0 web fetches and just hallucinated 17K very wrong tokens. Then the main agent decided to pretend to implement tp, and just copied the entire model to each node...

I guess our conscience of OpenAI working with the Department of War has an expiry date of 6 weeks.

Thing is that Anthropic was always working with DoD, too, and the line in the sand they drew looked really noble until I found it didn't not apply to me, a non-US citizen. Dario made it clear that was the case.

And so the difference, to me, was irrelevant. I'll buy based on value, and keep a poker in the fire of Chinese & European open weight models, as well.

Re: Claude Opus 4.7

#408

Post by j_bum »

bayesnet wrote: post #276

This is a CC harness thing than a model thing but the "new" thinking messages ('hmm...', 'this one needs a moment...') are extraordinarily irritating. They're both entirely uninformative and strictly worse than a spinner. On my workflows CC often spends up to an hour thinking (which is fine if the result is good) and seeing these messages does not build confidence.

Agreed. I actually have thought those were “waiting to get a response from the API” rather than “the model is still thinking” messages

Re: Claude Opus 4.7

#409

Post by ctoth »

cupofjoakim wrote: post #20

> Opus 4.7 uses an updated tokenizer that improves how the model processes text. The tradeoff is that the same input can map to more tokens—roughly 1.0–1.35× depending on the content type. caveman[0] is becoming more relevant by the day. I already enjoy reading its output more than vanilla so suits me well. [0] https://github.com/JuliusBrussee/caveman/tree/main

1.35 times! For Input! For what kinds of tokens precisely? Programming? Unicode? If they seriously increased token usage by 35% for typical tasks this is gonna be rough.
Post reply on HN