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

Why does Opus 5 feel worse to work with?

mun-logadan.github.io

Post reply on HN
Print view View original

241–250 of 915 posts

Re: Why does Opus 5 feel worse to work with?

#241

Post by loopmonster »

stingraycharles wrote: post #226
barrkel wrote: post #12

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…

> Sentences that orbit a point, then jump to it like it's a revealed insight. Is this inside the thinking tokens, or the output? As this type of stuff is expected for thinking, because of the whole CoT / “think step by step” works, as this is optimal for the way LLMs work with attention and next word prediction. So the fact that it first “orbits” a point only to get to the conclusion afterwards is the system working…

It's the output.

Re: Why does Opus 5 feel worse to work with?

#242

Post by ramon156 »

I've been using deepseek + GLM for a week now (i haven't even hit $10 yet, excluding the GLM subscription which is already paid for).

Then I asked Opus 5 to do an analysis of the new code, docs, tooling, and tell me what it finds.

It found "six bugs", made an artifact of it (not sure why) and then fixed said bugs. two out of those were unfinished tasks. They weren't bugs yet per se, think of a prefix that wasn't setup for an object that was unused anyway.

The other four were not bugs and it just updated documentation along with a "regression prevention test". It wasn't a bad suggestion, but calling it a bug was odd, and I'm unsure if this was going to be an issue regardless as it was documented somewhere else.

Anyway, I already hit my session limit after this, so deepseek and glm are grinding away again, doing more progress than claude does in the 40 minutes it takes to analyse code.

I'm glad claude is shipping auto-mode. I hope OpenCode integrates something similar soon.

Re: Why does Opus 5 feel worse to work with?

#243

Post by elromulous »

gwbas1c wrote: post #239

When I tried Claude 5 (Fable?) (In Visual Studio via Copilot,) the results weren't as bad as a lot of the comments here... But it was super-slow. IE, so slow that I could code faster than it, negating the entire point of using AI to begin with! I went back to Opus 4.8, but recently switched to GPT 5.6 Luna. The results are comparable in quality, but it's much cheaper and much faster. --- The thing with coding agents…

Regardless of the underlying issue, I don't agree with:

> negating the entire point of using AI to begin with

I can almost certainly wash dishes faster than my dishwasher, but the dishwasher frees me up to do other things. Not to mention, you can run many agents in parallel. Resulting in your overall code production throughput (the issue you're raising) being greater.

Re: Why does Opus 5 feel worse to work with?

#244

Post by jbs789 »

MEMORYC_RRUPTED wrote: post #13

It's not even code for me, but the prose it writes. For some reason, the way Opus 5 "talk" elicits frustration in a way that 4.5 to 4.8 never did. Can't put my finger on why, but I've flipped over to Codex because what it produced wasn't worth the frustration.

I found myself swearing at it more recently.

It was going off today about having “shipped” something and I was like no… nothing has even been committed.

And then it produced an incredibly verbose comment about hypothetical future changes. And all I could think was sure, let’s keep it short, or add a simple test that will break if that hypothetical becomes true.

Or maybe I’m just more easily annoyed recently…

Re: Why does Opus 5 feel worse to work with?

#246

Post by jbs789 »

roarcher wrote: post #55
cyberrock wrote: post #32

Small specific complaint: whoever is making Opus love using git checkout to mutate test, please stop. IME it's a footgun that it shoots itself with every single day. I'd rather it pollute git stash than watch it git checkout and forget the reverted file.

I explicitly forbid Claude to make any changes to Git state in my global CLAUDE.md, but every so often if I let it perform a task in Auto mode, after it finishes it will remorsefully confess to having used git checkout to test a change. I suppose that its RLHF training has taught it that asking forgiveness later is sometimes a useful workaround for annoying restrictions.

I noticed it using git more too and it was frustrating me bc I view that as stepping too far. It does seem something has changed.

Re: Why does Opus 5 feel worse to work with?

#247

Post by ryanbrunner »

netniuq wrote: post #87

Opus 5 feels like it's meant to be used as a very focused subagent under Fable, not user-facing. Which is a downgrade to what Opus used to be, but would imo absolutely have made sense for Anthropic when you consider that we all should have been paying API pricing for Fable in Anthropic's original plan. From the capabilities side it's similar, so it's basically just an upsell to Fable 5 if you want to keep your sanity…

I find the other direction works ok as well - Sonnet 5 with Opus as an advisor - all the "Opusisms" get hidden from you since the convo is between Sonnet and Opus but you still get pretty decent results.

Re: Why does Opus 5 feel worse to work with?

#248

Post by strulovich »

stefan_ wrote: post #222
inferniac wrote: post #207

Earlier quoted context omitted.

the tip that was floating around on x was to tell it to use "ASD-STE100 Simplified Technical English" cladue desktop has an instructions sections under general options, you can put something like "try to stick to ASD-STE100 Simplified Technical English, keep answers short and to the point" funnily enough the placeholder they suggest when its empty is "keep answers short and to the point"

CLAUDE.md is mostly powerless against the reinforcement learned crap. I'm up to three separate instructions telling it to cut out the hyper verbose, retelling history comments and it still writes them every time.

The best trick I have after asking it nicely in all sort of ways is:

1. Have it build a scoring script that penalizes words outside a simple English list and approved jargon. Penalize sentences over 15 words as well. Add whatever else.

2. Run it in a loop to reduce the score while preserving intention

This works much better than other ways I’ve tried. Of course it costs more. And I would apply it only to the output to the user, not the thinking process (I think the AI thinks better with their crazy English)

Of course, sometimes nuance is lost by this process. That’s just the nature of making things simpler.

Re: Why does Opus 5 feel worse to work with?

#249

Post by jbs789 »

taspeotis wrote: post #148

My completely unfounded pet theory is that it’s been ruined by the masses. Claude Code in the hands of normies spamming “3” and “y” to send their transcripts to Anthropic. Rated “3” simply because they are not software developers and are just amazed at what Claude has built visually, not technically. And thus the training has been poisoned.

That’s fascinating. You could be on to something.

Re: Why does Opus 5 feel worse to work with?

#250

Post by Laurel1234 »

disgruntledphd2 wrote: post #155
D13Fd wrote: post #120

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”…

Yeah, I actually have started using GPT Sol much much more, as Claude (all of them) were far too trigger happy around making changes, and refused to listen to my requests to take things slowly. 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…

I got the same impression. It feels like a massive misstep too; it makes sense to have models tuned for this but they should definitely be separate from those aimed at software engineers.
Post reply on HN