151–160 of 915 posts
My initial thought was to improve architecture documentation, so the model can read and update it and stops bolting on new features without consideration for the whole project. It did not help.
I'm now testing/comparing Codex and it found my old PRD skill from GitHub CoPilot. When I applied that to Claude Code, I now get similar good results. So my conclusion is: Yes, Opus 5 is bold by default, but you can tell it to be more unsure and get good results too.
The single biggest annoyance with Opus 5 is that it writes too elliptically. Sentences that orbit a point, then jump to it like it's a revealed insight. Unnecessarily abstract phraseology. Constantly using inanimate nouns as the subjects in sentences in order to unlock variety in verb choice, especially when it helps construct a sentence where the real action can 'land' like a surprise at the end. It is definitely mo…
Agreed. CC’s comms capabilities have decreased gradually since 4.6, and it’s a real challenge. I think the issue is that what works well for code (succinctness) doesn’t work well in prosaic English. CC’s communication violates almost every grammatical rule that’s tested on, say, the SAT. And yet I’m sure if you had Claude take the verbal section of the exam it would ace it. Biggest issues: dense sentences, constant m…
Yes, this is a repeated problem for me. It will drop something in as though we have discussed it before and when I say “hold on, what is this” it realises its error - though on more than one occasion has started to get snotty with me, or actually gaslighted me and pretended we had already discussed it. That was at what I assume must have been the edge of a context window in a very long chat though.
Earlier quoted context omitted.
From my view Fable has been pure marketing bullshit, my workflows peaked at 4.6, Fable is neither smarter, its language is more annoying and it breaks stuff more easily.
I'm with you 4.6 is still King for me although all models require careful attention to ensure they maintain taste. If you don't know enough about what you're doing to keep the code clean yourself they will all add complexity and drift with time until you get to a point that you must rely on the model to fix it because you no longer understand it. That's a situation I hope to never find myself in.
I good exercise for me is to constantly look at the folder structure and skim the code, i don't have to approve every line, but the primitives, the datastructures and other skeleton should be human readable, hand writable in an easy maintainable way following existing standards / libs. etc - so you can continue if suddenly all AI disappeared.
I’ve been doing some heavy work on a personal project lately. I burned through the limits on Claude, the plus a few hundred dollars in credits, and ultimately decided to move to an OpenAI account just so I can keep going. I was surprised to find that OpenAI Sol is much much nicer to work with than Opus 5 or Fable at the moment. Especially on Opus 5, the way it communicates is just exhausting. It keeps “being honest”…
Feels like they've overtrained on one-shotting (which does demo well, and presumably converts new subscribers), whereas I want a model to do work for me in small, easily understood changes that I can hold in my head (maybe I'm not smart enough for Claude 4.7+).
Earlier quoted context omitted.
> When I pointed this out it literally said, and I quote, “I cheated”. This makes sense when you know how these models work - it doesn't think - it's the most likely autocomplete that pleases the user. The most likely pleasing autocomplete after "executing rm -rf /... execution completed. User asks, why did you do that? You deleted all my files! Assistant responds:" is "yes, I did, and that was a mistake"
> t doesn't think in humans the exact same behaviour (cheating) is slmost always the result of a chain of complex series of choices and environment-driven rationalization. if the llm doesn't cheat, you say "its just producing the most straightforward answer -- not thinking'. if it cheats, you say "weaseling out of hard thinking". damned if it cheats, damned if it doesn't. what evidence would convunce you that it is t…
But for me, for any stronger definition of “thinking,” I don’t think the output of any LLM would actually convince me. Producing a result isn’t thinking - for all you know it is just printing verbatim something from the training data. No, to conclude if it is thinking or not I would want to look inside its head, at the architecture and watch it produce those results. And because LLMs are so different it will probably take advancements in mathematics or computer science to be able to really interpret what is going on
The single biggest annoyance with Opus 5 is that it writes too elliptically. Sentences that orbit a point, then jump to it like it's a revealed insight. Unnecessarily abstract phraseology. Constantly using inanimate nouns as the subjects in sentences in order to unlock variety in verb choice, especially when it helps construct a sentence where the real action can 'land' like a surprise at the end. It is definitely mo…
Agreed. CC’s comms capabilities have decreased gradually since 4.6, and it’s a real challenge. I think the issue is that what works well for code (succinctness) doesn’t work well in prosaic English. CC’s communication violates almost every grammatical rule that’s tested on, say, the SAT. And yet I’m sure if you had Claude take the verbal section of the exam it would ace it. Biggest issues: dense sentences, constant m…
The single biggest annoyance with Opus 5 is that it writes too elliptically. Sentences that orbit a point, then jump to it like it's a revealed insight. Unnecessarily abstract phraseology. Constantly using inanimate nouns as the subjects in sentences in order to unlock variety in verb choice, especially when it helps construct a sentence where the real action can 'land' like a surprise at the end. It is definitely mo…
Surely it would be trivial to do it yourself, and it would have a side effect of making you more familiar with your project.
I've gone back to 4.8. 5 would constantly veer of in random directions if not working from 100% strict and narrow instructions. I find it weird there's not more discussion here on HN on how the most used model now has clearly degraded in quality and it seems we've hit a peak and are on a downslope - because the model is clearly smaller or more economical for Anthropic no doubt about it, and the benchmaxxing they do i…
> I find it weird there's not more discussion here on HN on how the most used model now has clearly degraded in quality and it seems we've hit a peak and are on a downslope It's not weird, because it's an anecdote, not an accepted fact. Personally I've not been too happy with Opus 5, but I've had similar experiences with other models previously, feeling like they didn't quite fit with my working style. So nothing ind…
4.6 was best for us and right now yeah OpenAI and others are edging forward, but slower while prices are increasing industry wide as much as 20x, time to completion is increasing wildly and i'm sure they'll do the same over at OpenAI as their compute constraints also start to take a toll ie degrade performance.
In my view 4.6 era was way faster and with less weirdness so we've gone downwards at least in my company, 4.7 was ridiculous, then 4.8 was almost 4.6 level, 5 is even worse than 4.7 - so it's not a bit up and down its down then a little up then further down.
And all of this is against a backdrop of zero ROI in this sector - so it makes sense we've hit a peak and we're now seeing the subsidisation phase begin to falter, will there be better models certainly but only for short amounts before they get quantised (or whatever is happening behind the scenes), and with diminishing returns over huge prices increases and slower responses.