Hacker Newsnew | past | comments | ask | show | jobs | submit | fancyfredbot's commentslogin

Has anyone used qwen 3.5 9b? Did it really compare well to Opus 4.1 outside of synthetic benchmarks?

In September 2025, Executive Order 14347 authorized the usage of "Department of War" as a secondary name, which is now preferred by the department.

Cool

If you'd bought every fad coin your brother recommended to you, even though it was a bad idea in your own judgement then I think you'd have zero dollars.

My brother is not a crypto guru, we just discussed about BTC. It was a philosophical conversation in the spirit of perhaps it will become an alternative to conventional money, then it could greatly increase in value.

But I agree that buying new coins is like buying lottery tickets - if you buy all the tickets, there will be no money left.


This isn't a particularly acute or interesting comment but I feel the need to say: This is a fantastic, well written, and quite sympathetic account of the excesses of the world silicon valley VC has created. It's weirdly beautiful.


Simple prompts which illicit incorrect responses from recent LLMs will get you on the front page of HN.

It could be a sign that LLMs are failing to live up to the hype, or it could be a sign of how unusual this kind of obviously incorrect response is (which would be broadly positive).


It's all trade offs. The router works most of the time so most free users get the expensive model when necessary.

They lost x% of customers and cut costs by y%. I bet y is lots bigger than x.


I have a feeling the days of patent encumbered video codecs will come to an end soon and be replaced with some kind of autoencoder, or at least the decoder part. It should be possible to match or exceed the compression achieved by H.265, although the decoder would probably consume more energy and cost more. The cool thing about autoencoder compression is that at high compression rates it'll still look like a high resolution image, it'll just be of the wrong scene!


> I have a feeling the days of patent encumbered video codecs will come to an end soon

I have a feeling that the reason we get a new codec every couple of years is: money.


> You get AI that can make you like 90% of a thing! 90% is a lot. Will you care about the last 10%? I'm terrified that you won't.

Based on the Adobe stock price the market thinks AI slop software will be good enough for about 20% of Adobe users (or Adobe will need to make its software 20% cheaper, or most likely somewhere between).

Interestingly workday, which is possibly slightly simpler software more easily replicable using coding agents is about the same (down 26%).


The bear case for Workday is not that it gets replicated as slop, but that its “user base” becomes dominated by agents.

Agents don’t care about any of Workday’s value-adds: Customizable workflows, “intuitive” experiences, a decent mobile app. Agents are happy to write SQL against a few boring databases.


Is it still a decent gamble after you've been trying (and failing) for a decade, and numerous well funded competitors are going the easy way, and when there is huge upside to being first, and when the value of FSD easily covers the rapidly falling cost of LIDAR?

No. It's not a good idea. It's not a good gamble. It's stupid, and the engineers can see it's stupid. A lot of them have quit, reducing the very slim chances of it working even further.


But why is FSD "failing" is the key question.

Hint: it's not the sensor inputs that are the bottleneck!


Yeah you could be right.

Not my area of expertise, so I’d rather not try to predict what will work and what won’t.


'You have to help us as local car mechanics drive the race car!'


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: