frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Claude: Degraded Performance for Multiple Models

https://status.claude.com/incidents/q7txxvbsftgq
88•matt89•38m ago

Comments

bayganyo•29m ago
Here we go again...
bulverismo2•27m ago
ok, i am not crazy
ray_v•24m ago
well, I wouldn't go that far .. but in this small, narrow case ... no.
worldsavior•6m ago
You're saying he's crazy.
carterschonwald•27m ago
ive found degraded performance on models larger than 4.7. i assume its model damage from overly self righteous post training resulting in false/feigned balance imported into any long running complex task.

wish i was joking.

brcmthrowaway•24m ago
Aren't the model weights frozen?
mceachen•20m ago
Model competence is an interaction of weights, system prompt, and harness.
kardianos•23m ago
I've switched off claude this week; the last week has been significantly degraded in ability, many more screw-ups.
hinkley•26m ago
I wonder if they’ll ever find that someone has tricked the models into doing work off the books. If they did the incident report might look like this, especially if someone got greedy instead of keeping it small. Or screwed up.
isoprophlex•25m ago
With the Opus models spouting more and more gibberish as version numbers increase, the joke about what "degraded performance" means basically makes itself
swader999•24m ago
And we get our subscription usage cut in half tomorrow if I remember correctly? EDIT: By a third. Thx below.
saaaaaam•23m ago
What?!
eamag•22m ago
by a third (it was 50% increased)
birdman3131•21m ago
cowork was 100%
echelon•21m ago
Open source, here I come.
ramoz•19m ago
Source required here
swader999•17m ago
https://usingclaude.com/en/news/updates/claude-code-weekly-l...
saaaaaam•23m ago
This feels like a near daily occurrence.
gaigalas•23m ago
This age: we made the thing that codes faster before we made the thing that does QA faster.
__MatrixMan__•21m ago
Nothing new here. Except for the most trivial of bugs, finding and reliably replicating the bug is almost always harder than fixing it.
gaigalas•14m ago
I lived in a short period of time in which QA was really good. Early Jenkins era, before GitHub. People engineered a lot of ingenious stuff to prevent bugs.

One team I worked with had tests for the product we made ranging from IE6 to IE11, for example. We did demos in-company where people would poke at the products before launch, play with it. When it reached production, it was rock solid stuff. Our motto was "quality is non-negotiable": we were willing to cut scope but never rush things.

I think things changed since then. "Move fast and break things" was a change, and the bill always comes.

drums8787•23m ago
Our week of discontent.
gzer0•21m ago
Nooooo I'm going to have to use my brain again and write 100% of my code like a caveman from December 2024.
rvz•21m ago
Claude is taking a watercooler break for now. Just like a human would.
sreekanth850•20m ago
Anthropic had really screwed up after 4.6. i don't know if they work to satisfy the ego of themselves or for releasing a better model for tasks.
CSMastermind•15m ago
After using Fable more extensively, I've found that it often is lazy or lies or tries to take shortcuts. For a company so sanctimonious about alignment, they seem to be the ones doing the worst at it.

Availability aside they've really made me appreciate OpenAI and cheer for other competitors in the marketplace even if I have mixed feelings about using Chinese models.

fellowniusmonk•10m ago
There are whole sections of code work that 4.7+ can't do simply because it is both over fit and stubborn.

God save you if you have a company with narrow but correct technical tradeoffs, because you operate at scale.

Opus from 4.7 one will wreck your code and argue for hours with your engineers.

Certain parts of our company have had to mandate 4.6 and a training doc to explain why our current choice is both the cost efficient and performant one and shouldn't just be ripped out.

Newer models will re-litigate the same bad, known failed architectures over and over again.

baxtr•4m ago
Just recently went back to ChatGPT after abandoning it for Claude. I must say I was stunned at how good it had become and also how they introduced new product features that I really liked.

I wonder if from now on we have to switch providers every six months or so.

taytus•3m ago
fny•19m ago
Despite the years-long moaning on HN about AWS US East being a single point of failure, we've sold our souls to yet another unstable monolith.
echelon•14m ago
LLMs for coding are new. There are lots of alternatives, and there's a burgeoning open source compliment.

We'll be fine no matter how Anthropic fares.

lta•5m ago
Nobody forced you to sell your soul. You made a pact with the devil. We all know how this ends up
paxys•12m ago
Must be a day ending in Y
slimscsi•11m ago
Its called Opus 5
hmokiguess•11m ago
Mondays are for GitHub, Tuesdays are for Anthropic
corvad•6m ago
Wonder what Wednesday will be.
leumon•5m ago
They actually also had some issues yesterday: https://status.claude.com/incidents/zhk4v3yv1lsf
jmkni•4m ago
And github today lol https://www.githubstatus.com/incidents/bmpybhnrky3x
baxtr•2m ago
The AI apocalypse will definitely happen on a Monday. Remember, robots - unlike lazy humans - work weekends too!
i_idiot•5m ago
What's the incentive to keep on improving the model beyond a point? 10 devs on a team will be cut to 2 devs, so that's 8 licenses lost. They have to increase the price many fold.
chrisjj•5m ago
[delayed]
hirvi74•3m ago
While ancedata does not mean much, I have had horrible success with Claude lately. I have been using Claude to crosscheck some of the outputs from GPT and vice versa. It appears both Claude and GPT believe GPT's solutions are better (and so I do).

I still believe Claude has a better UI/UX in the web interface, but tolerating Anthropic's bullshit is not worth it.

magic_hamster•2m ago
To be honest, running Deepseek v4 flash 0731 is enough for most what I need, and I like its responses way more. It's crazy that I can run this in a Q8 quantization in a home setup. It feels and performs like a frontier model.

The only issue with relying on local models is when you need them to prompt other models, and you minght need to offload or switch models constantly which adds significant overhead.

But when it all works, its truly awe inspiring.

taytus•2m ago
Hopefully, a reset is coming.
bmulholland•17m ago
https://support.claude.com/en/articles/15910845-claude-code-...
scottg489•6m ago
Maybe I'm missing something here, but it sounds like limits were increased and now they're just going back to the levels they were at before?
aweiher•10m ago
Fable 5 promo https://support.claude.com/en/articles/15862783-claude-fable...
no_no_no_yes•17m ago
I completely forgot about this, I'm already budgeting usage until my reset haha
Every 3 months*

Claude: Degraded Performance for Multiple Models

https://status.claude.com/incidents/q7txxvbsftgq
90•matt89•38m ago•62 comments

Using the railway network as a flatbed scanner

https://philo.gay/linecam/
252•otherayden•4h ago•42 comments

The Amazon Tax

https://seths.blog/2026/08/the-amazon-tax/
425•herbertl•3h ago•298 comments

How I Under-Engineered My Book

https://chriskiehl.com/article/how-i-under-engineered-my-book
49•goostavos•21h ago•13 comments

Linux 7.3 improves performance when running out of vRAM

https://pixelcluster.dev/VRAM-Overcommit/
407•flaburgan•9h ago•161 comments

Fixing a Bricked Framework Laptop

https://quantum5.ca/2026/08/16/fixing-bricked-amd-7040-series-framework-13-laptop-with-20-tools/
185•jp_sc•3h ago•109 comments

Python Polars Cheatsheet (based on our O'Reilly book)

https://opensource.posit.co/resources/cheatsheets/polars/
79•jeroenjanssens•3h ago•12 comments

Cursor launches Origin, GitHub alternative

https://cursor.com/changelog/origin-code-hosting
156•tomasreimers•1d ago•104 comments

Launch HN: machine0 (YC S26) – Persistent CPU and GPU VMs from the CLI

https://machine0.io
6•bwm•40m ago•6 comments

Teaching my kid to code with a modern MUD

https://tau.dev/2026/08/07/canon
146•andrewjanke•6d ago•48 comments

Code-native generation of highly programmable 3D assets (2026)

https://arxiv.org/abs/2607.22738
22•baigy•2h ago•20 comments

Splitting a Git Commit

https://blog.gnoack.org/post/git-history-split
45•signa11•3d ago•18 comments

Composable Tests

https://newsletter.kentbeck.com/p/composable-tests
46•vinipolicena•3h ago•36 comments

Fairphone is now officially available in the United States

https://www.fairphone.com/nl/stories/the-fairphone-gen-6-is-all-about-giving-you-more
234•Vinnl•4h ago•117 comments

Babies born under sugar rationing grew into adults with lower cancer risk

https://theconversation.com/babies-born-under-sugar-rationing-grew-into-adults-with-lower-cancer-...
111•zeristor•3h ago•31 comments

Memory prices climb 500% in 12 months

https://www.tomshardware.com/pc-components/ram/memory-prices-climb-500-percent-in-12-months-up-to...
137•haunter•23h ago•99 comments

Google has acquired the data of failed US airline Spirit

https://www.theregister.com/ai-and-ml/2026/08/18/google-buys-crashed-airline-spirits-data-at-auct...
454•pseudolus•6h ago•324 comments

Meta Files Patent for Facial Recognition, Automatic Recording of People

https://www.privacyguides.org/news/2026/08/17/meta-files-patent-for-facial-recognition-automatic-...
178•DeepLogin•4h ago•119 comments

Superpowers, Not Superintelligence

https://bond.now/news/superpowers-not-superintelligence
3•edbernays•39m ago•1 comments

How Bluesky draws its logo on screenshots

https://timmarinin.net/2026/bluesky-screenshots/
664•gavide•18h ago•418 comments

Apple announces changes for apps in the European Union

https://www.apple.com/newsroom/2026/08/apple-announces-changes-for-apps-in-the-european-union/
63•newusertoday•45m ago•65 comments

Show HN: Shoehorn – Quantize any model down to run on your machine

https://notactuallytreyanastasio.github.io/shoehorn/
18•rhgraysonii•2h ago•2 comments

California's new tire efficiency rules could save drivers $1B a year

https://grist.org/transportation/californias-new-tire-efficiency-rules-could-save-drivers-1b-a-year/
53•littlexsparkee•14h ago•66 comments

Rethinking Database Programming

https://acadia.engineering/blog/rethinking-database-programming
179•honungsburk•9h ago•96 comments

Show HN: Openleetcode – local LeetCode runner where tests live in the repo

https://github.com/therepanic/openleetcode
5•therepanic•1h ago•1 comments

Finger: Social network that never died

https://en.andros.dev/blog/54572bc7/finger-the-1971-social-network-that-never-died/
129•andros•9h ago•40 comments

Quake Shareware, a CD-ROM just a little too full

https://fabiensanglard.net/quake_shareware_cd/index.html
458•shdon•19h ago•197 comments

Baking a Model: A Metaphor for LLM Training

https://newsletter.kentbeck.com/p/baking-a-model
27•KentBeck•3d ago•3 comments

Ask HN: Ive quit six systems for tracking my illness. What works?

12•Abh1Works•2d ago•10 comments

Show HN: Saggar, a Mac terminal that keeps sessions and your attention organized

https://saggar.marginalutility.dev/
43•mcclowes•1d ago•48 comments