frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: Why are Claude models so verbose?

1•JacobWolf•58m ago
Hello! First time poster, longtime reader. I’m a professional software engineer at a startup with about 50 engineers. My primary languages are Ruby (Rails), TypeScript (React), and Go.

Of my colleagues, I’m one of the ones who tries to keep up with new AI models on every release. When there’s a new frontier or open-weight model, I’ll give it an honest try for a few days as my main driver and report back to my colleagues.

I’m a Cursor user at work and in my side projects. My go-to models are GPT-5.6 Sol on high/xhigh for planning, GPT-5.6 Sol on low/Grok 4.6 on medium for implementation, and the occasional Claude Fable 5 for complex, harder-to-understand and implement tasks. Many of my colleagues are exclusively Claude users and use Claude Code.

I’m not the first one to observe this but I feel like any time I use Fable, Opus, or Sonnet, they are extremely verbose. They make a ton of extra code comments, they ignore instructions, and they invent solutions for things solved by language built-ins or preinstalled libraries like Active Support and es-toolkit. I don’t see this happening with any other model at the same rate.

My question are: Why do we think the Claude models do this at a seemingly higher rate? And have people found any effective means to curb this behavior?

Comments

danielovichdk•51m ago
Because models that chat to themselves generate more tokens, hence more money is spent.

Easy

onion2k•30m ago
That's an over-simplification though, because models that waste tokens will lose users. Efficiency is a great marketing feature.
thingstohappy•39m ago
I think there are a few nuances here:

1. On the ignore instructions point: using Cursor with Claude as the underlying model is quite different from using Claude directly through Claude Code. The agent/tooling layer can have a pretty big impact on the behavior.

2. On verbosity: IMO, Claude's default writing style tends to be more verbose than GPT's. GPT models are generally more concise when coding, while Claude often explains more. That said, I think this can be mitigated quite a bit by how you harness the model.

3. I agree that it’s worth trying new models. The pace of AI development is so fast that models can change significantly in a short time. But I’d also suggest trying different agent tools rather than sticking with Cursor alone. The tooling and agent loop can make a big difference too.

JacobWolf•4m ago
I churned off of Cursor for about a year to use Claude Code from spring 2025 to spring 2026 and I noticed it’d regularly ignore CLAUDE.md & Claude rules. It’s part of why I came back to Cursor. The Cursor harness has produced more reliable results, but the excessive verbosity is still pretty common across harnesses, in my experience.

Someone just published open knowledge base

https://github.com/mo-root/open-kb
1•moroot•1m ago•0 comments

Two unvaccinated people in Pennsylvania are first measles-related 2026 US deaths

https://www.cnn.com/2026/08/25/health/measles-deaths-pennsylvania
2•helsinkiandrew•4m ago•0 comments

Nvidia Jetson Orin-guided Russian AI drone killed three civilians in Ukraine

https://www.tomshardware.com/tech-industry/drones/nvidia-jetson-orin-guided-the-russian-ai-drone-...
2•sbulaev•6m ago•0 comments

Apple Mac Studio M5 Ultra with 1.2TB/S Memory Bandwidth

https://www.apple.com/mac-studio/
1•itchingsphynx•7m ago•0 comments

Another DDoS attack against Norwegian government IT infrastructure – status

https://status.digdir.no/incidents/d7tgwqgzd742
2•omBratteng•10m ago•0 comments

Show HN: Launched free browser agent (the catch: ads)

https://rtrvr.ai
1•arjunchint•10m ago•0 comments

Can AI Music Tools Replace Epidemic Sound? An Honest Look

https://www.musegen.ai/blog/ai-music-vs-epidemic-sound
1•musegenai01•11m ago•0 comments

Show HN: Multiplayer Laser Simulator

https://prismaduel.com/
1•maxberggren•15m ago•1 comments

Chinese government fears talking to AI might replace human intimacy

https://www.theguardian.com/world/2026/aug/26/china-ai-companion-relationships-marriage-birth-rat...
1•fittingopposite•15m ago•2 comments

Turn Any Image into JPEG XL. Free and Private

https://anyjxl.com/
1•johnnyApplePRNG•18m ago•0 comments

GitHub denies Copilot Autofix wrote the bug Wiz's AI agent found

https://ai-news.ghost.io/github-denies-copilot-autofix-wrote-the-bug-wizs-ai-agent-found/
1•Striation•19m ago•0 comments

LLMs prefer machine-written prose over Austen and Shakespeare ~90% of the time

https://shivanshuag.com/blog/the-machine-never-raises-its-voice/
1•shivanshuag•23m ago•0 comments

More than half of adults in U.S. say they lack basic statistical understanding

https://www.psu.edu/news/research/story/more-half-adults-us-say-they-lack-basic-statistical-under...
4•giuliomagnifico•23m ago•0 comments

Run several Linux kernels at the same time on bare metal, without a hypervisor

https://lwn.net/ml/all/ao34RJ7aZ2BLd67S@pop-os.localdomain/
3•NewJazz•24m ago•0 comments

Why Political Moderates Are Losing

https://www.thebignewsletter.com/p/why-political-moderates-are-losing
2•connor11528•30m ago•0 comments

I think printable objects are an underrated way to test physical products

https://deskidols.vercel.app/
1•Shmungus•30m ago•0 comments

First Findings of Mirror-Like Activity in Individual Human Neurons

https://www.caltech.edu/about/news/first-findings-of-mirror-like-activity-in-individual-human-neu...
2•XzetaU8•36m ago•0 comments

DefCon 34 - Stalking the Wily Hacker: 40 years later – Cliff Stoll [video]

https://www.youtube.com/watch?v=656058JxTM0
2•wizardforhire•41m ago•0 comments

Iran war supercharged EV sales everywhere except the U.S.

https://yaleclimateconnections.org/2026/08/trumps-iran-war-supercharged-ev-sales-everywhere-excep...
1•toomuchtodo•43m ago•1 comments

We run Postgres on local NVMe

https://pgrun.dev/blog/why-we-run-postgres-on-local-nvme
2•handfuloflight•44m ago•0 comments

Lab supply companies have been selling antibodies using manipulated images

https://arstechnica.com/science/2026/08/lab-supply-companies-have-been-selling-antibodies-using-m...
1•OutOfHere•46m ago•0 comments

Fetch – control your Claude Code agents from the Mac notch

https://tryfetch.dev/
1•laithistaken•47m ago•0 comments

Recursive Self-Improvement (not yet). Also, we need a new definition

https://www.youtube.com/watch?v=VQanDOFJwvo
1•bradwmorris•48m ago•1 comments

Goodbye File Browser

https://hacdias.com/2026/07/28/filebrowser/
1•Ginop•48m ago•0 comments

Ask HN: Why are Claude models so verbose?

1•JacobWolf•58m ago•4 comments

Xiaomi's Xring Series suggests more OEM's switching to in-house SoCs

1•StizzurpXDD•58m ago•0 comments

Perplexity's Portable Computer tackling local AI services market

https://www.theregister.com/ai-and-ml/2026/08/26/now-perplexity-is-trying-to-get-into-the-local-a...
1•Gaishan•59m ago•0 comments

Running out of money': Kraft, McDonald's, Whirlpool CEOs flag consumer concern

https://finance.yahoo.com/economy/articles/running-money-kraft-mcdonald-whirlpool-114500035.html
8•MrJagil•59m ago•1 comments

BanglaTools – Bengali Unicode and Bijoy text conversion tools

https://banglatools.com/
1•btref26hn•1h ago•0 comments

Front end skill pack for AI agents, with machine-enforced quality gates

https://krishna-modi12.github.io/frontend-design-pro/
1•KrishnaModi12•1h ago•1 comments