News for hackers. Stuff that matters.
Re:Anthropic hits $2T in Nasdaq debut by Tajnymag on Thursday November 19, 2026 @03:27PM (#50565897)
I swear he got so much dumber in the last two weeks. Today, Opus acted downright lobotomized in certain queries, not following simple commands.
Re:Anthropic hits $2T in Nasdaq debut by bluegatty on Thursday November 19, 2026 @03:36PM (#50565998)
Dude. I hear you. I don't like to be a conspiracy theorist - and I'm pretty sure they don't materially change the models like that ...

... but *god dam* it feels like 2025 today.

I swear Ima about to throw my laptop out the window - it's been one of the most frustrating few days of my career.

I think this may be an ongoing problem: The AI is 'poweful but broken' - aka it does not feel like a working project by it's very nature.

It's like a fast car, with doors that won't close, gas pedal 1/2 works, brakes 1/2 work.

Is it faster than walking? Yes. But it's just crazy - and Im beginning to think this is just a property of the AI and that overcoming it woudl require something like 10x the compute .. which is 'not worth it'.

So - we won't get a smooth ride for maybe 25 years, when RAM/GPU become and absolute commodity, and the work we do now is goes through 10 filters/passes before you see an answer so its not 'full of **' 1/2 the time.

There is a new psychology here.

I love development, I love AI - but this is driving me completely crazy.

Re:Anthropic hits $2T in Nasdaq debut by hexad74 on Thursday November 19, 2026 @04:03PM (#50566333)
I always wonder if I am hallucinating and having a bad day or this is actually happening. Yesterday and today Opus 5.5 has been making some strange mistakes. I have a workflow for testing a series of jobs that run in SQL agent. Instead of just generating a script to test the SQL directly as we have done 100 times, it decided to cursor through job steps and attempt to inject the debug parameters into a SQL string pull from the job.

I have noticed a few other silly choices it as made as well.

Re:Anthropic hits $2T in Nasdaq debut by MisterMunchkin on Thursday November 19, 2026 @05:51PM (#50567664)
They quantise the model right before a new release so that the new model seems more intelligent.

In reality they probably don’t even have a new model, they just slightly up the quality of the existing model.

Re:Anthropic hits $2T in Nasdaq debut by bluegatty on Thursday November 19, 2026 @05:56PM (#50567733)
This is basically conspiratorial. You'd have to provide some evidence of this quantiazation
Re:Anthropic hits $2T in Nasdaq debut by system2 on Thursday November 19, 2026 @10:00PM (#50569624)
Absolutely not. Opus 4.8 was so op when it was released; I would launch a few projects a day. You can't even launch a project in 1 week with Opus 5.5
Re:Anthropic hits $2T in Nasdaq debut by troglodytetrain on Thursday November 19, 2026 @06:18PM (#50567968)
Yep. A few days ago I had fable build me a full adb bridge toolset and then go ahead and install an esim I said I had with a provider, it could check my gmail. Use xxx pin, etc. And it did it, quickly and without refusal. I was working during off hours for USA. Then today, big massive refusal increase combined with intelligence drain. It is 100% being done. You can even review the Claude subreddit with a graph of estimated performance that goes up on the weekends when no one is working and then falls during the week. I suspect at this point the security classifier is also being used as a 'oops its for safety' knob they can turn to effectively load-shed subscribe users based on need.

Ever notice how security refusal convieniently only either A. Give you a notification that pauses inference until you manually approve? Or B. Downgrade you to 'Opus 4.8' which is almost certainly just a relabelled haiku. Opus 4.8 was pretty damn good before, I spent a lot of time with it.

But service outages is bad for the ipo. So, better make sure it goes smooth. Eventually actual employees will tell us stories about this. But I'm sure that isn't going to happen till their shares divest.

Re:Anthropic hits $2T in Nasdaq debut by razster on Thursday November 19, 2026 @11:55PM (#50570338)
That is interesting, thanks for sharing.

Seems I picked the wrong day to come back to Claude and see if we wanted to try it after the hype. We primarily use DeepSeek/GLM for heavy programming. I gave Fable 5.2 some time this week, and lord, it was just bad. The formatting is what drove me away, along with the odd workarounds it performed on some of the tasks. I even had it screw up on an out-of-the-blue "ha ha" question: how many R's are in "Strawberry Rhubarb"? The SOB said 6. That right there I just ended my account, used the last of my tokens to do my weeklies, and closed the account. Sorry, but no thanks.