alt Hacker NewsDefinitely noticed some anecdotal declines this past week or so. Missing steps it didn't previously, needing reminders of tool calls and skills that it should auto apply.
Lots of "this is a potential issue, I didn't check into it though" which is super frustrating to see. The model just points out major flaws in logic or barriers but goes no further than a comment on it.
I've got a Qwen coder model and a flash next model locally now, so it's not the end of the world. But it does need to be remedied - are we seeing them prep for a new model deploy when this happens? Or are they experimenting with leaner hardware to drive down costs ahead of IPO?
I’ve been experiencing unreliable performance for a while and can’t imagine this is solely due to new model training.
It does seem like Anthropic is experimenting with how much less they can offer users. I also get the feeling they are judging what resources a task deserves and dumbing things down accordingly. I’m seeing lazy responses and skipped steps more frequently for non-technical queries.
It’s definitely possible the average user hasn’t noticed, unfortunately.