Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Hy3

hy.tencent.com

Post reply on HN
Print view View original

111–120 of 125 posts

Re: Hy3

#111

Post by jingpostmedia »

cbg0 wrote: post #20
minraws wrote: post #3

I tried out the model it's pretty great, better than ~~gpt5.4~~ gpt-5.4-mini perhaps, atleast close enough to sonnet 5 in performance that I didn't notice much of a gap. Not really at gpt 5.5 tier though, and probably below glm 5.2... But most of all it just works for me for most things I tried and it's exceedingly cheap so there is no reason not to use it, if you need a foss model. Edited: gpt-5.4-mini not the base…

Hy3 DeepSWE - 28% GPT5.4 xhigh DeepSWE - 52% A lot of contaminated benchmarks in the blog post about Hy3, needs real testing though I have a distinct feeling it's benchmaxxed like a lot of Chinese models.

[flagged]

Re: Hy3

#112

Post by flir »

pelagicAustral wrote: post #97
tarruda wrote: post #49

Earlier quoted context omitted.

Recently tried the pelican test on GPT-OSS which was probably one of the best local models of 2025. So cool to see how models have improved in the SVG pelican!

I'm skeptical, as tests go, I think that's burned out now. They could easily be training specifically to get a better pelican...

Doesn't have to be specific training, just has to consume simonw's blog. It's got lots of SVG pelicans, with helpful commentary on how good they are. I think there might be some kind of hill climbing going on here.

Re: Hy3

#113

Post by red-iron-pine »

redwood wrote: post #41
docheinestages wrote: post #39

What we really need is a breakthrough in inference or LLM architecture to allow running GLM-5.2-level models at the size of Qwen 3.6 27b or smaller on consumer devices like a 48GB Macbook Pro, and at least at 100 tokens/second. My hypothesis is that a smaller, less capable but faster model paired with a good harness can run for longer and brute force its way out to solve problems that the bigger models can one-shot.

That would be great during the winter months

used to heat my dorm room with a Pentium D that I overclocked and pointed a small fan at. Could open a window in winter, turn off the room radiator, and keep it cozy.

perhaps at-home LLMs will bring me back to that. fun days of hacking and thermodynamics.

Re: Hy3

#114

Post by numpad0 »

Azantys wrote: post #110
numpad0 wrote: post #109

Earlier quoted context omitted.

let me just say, you're not going to sound smart saying that

Useless comment

If you can't make yourself sound adequately smart, it could lead to people ignoring you, and/or acting in spite of your opinions/logic, and/or spending extra effort trying to decipher you. That is not an optimal situation, especially in cases where you would be right[1].

1: I don't think you're right in this instance, but that's beside the point.

Re: Hy3

#116

Post by righthand »

kordlessagain wrote: post #86
righthand wrote: post #43

Earlier quoted context omitted.

[flagged]

I will vouch for Simon. I do not know him personally. To me his posts are honest and inspirational. > I have been overly critical and arguing in bad faith about your writing in the past I think you are just critical without a stated valid reason. Arguing in bad faith seems to be a thing if HN history is the judge. And this is coming from a critical thinker, who is a bit tired of people firing off "human slop" comment…

That’s great for you and Simon but I personally don’t think there will be enough people in the future to use or appreciate these things that can be built and eventually no who will care to build them, and while this “discovery of LLMs phase” might be fun it diminishes the value I used to get from the job, which was writing code, problem solving, discussing and learning with people (having real people and mentors to work with and look up to), and just playing with computers.

There will be a few, sure but not enough. And even if there are enough to give me the things I miss, who is the expert? All of them with the exact same capability to, for example, generate some form of pelican drawing. C’est la vie, but it doesn’t mean it’s not depressing.

It has nothing to do with any attachment to a project or releasing something to everyone. It is a completely personal view about how the community and reward I got from it is dying. If you remove a technical abstraction layer you remove the technical abstractors.

There is a wide gulf of people who use LLMs and those who don’t and I didn’t fully realize this until the last few months. But it really is just a few industries making all the noise.

Re: Hy3

#117

Post by righthand »

simonw wrote: post #33

Pelican from a few days ago: https://simonwillison.net/2026/Jul/6/hy3/ - I was using the free tier on OpenRouter, which expires on July 21st. I tried the preview model 41 days ago and got a pelican with a "change pelican color" button: https://static.simonwillison.net/static/2026/hy3-preview-pel...

I have been overly critical and arguing in bad faith about your writing in the past. As well as negative towards you, which in turn was breeding a bad environment. While I dont really enjoy LLMs, you did help me realize my unreasonable feelings as well as realize the occupation (and the joys I got from it) is essentially dead from it’s previous iteration and that I should let go and just join in the “I’m doing it for the money and attention” crowd. I will still just hand code my own projects and not use LLMs when I can. I think it’s cool you started the pelican meme however useful it really is even if only aesthetically.

Re: Hy3

#118

Post by lostmsu »

ckocagil wrote: post #36
UncleOxidant wrote: post #16

Earlier quoted context omitted.

That's a 2-bit quant of DS4 flash. You're probably better off running Qwen3.6-27B at Q8.

Isn't Q8 way overkill these days? I see many graphs showing Q4 or Q5 having less than %1 deviation. Nvidia's NVFP4 Qwen quantization should be even better due to its better training methods.

1.01 over 30k tokens is over a googol (a large number with 100 zeroes)

Re: Hy3

#119

Post by nozzlegear »

Azantys wrote: post #103
fc417fc802 wrote: post #74

Earlier quoted context omitted.

> with barely any logic or reasoning I take it you enjoy works of literature with inconsistent world building? Or do you mean professional as opposed to creative writing? Because the bar is even higher for that.

The wording wasnt very good I ment compared to programming or math the amount of logic and reasoning is small (Research level math hardly compares to writing a book in raw reasoning and logic). And I thing the smaller models have enough "intelligence" to write coherent with logical world building, but only the big models can truly do hard math and programming work

> The wording wasnt very good

Writing isn't so easy after all.

Re: Hy3

#120

Post by kordlessagain »

righthand wrote: post #116
kordlessagain wrote: post #86

Earlier quoted context omitted.

I will vouch for Simon. I do not know him personally. To me his posts are honest and inspirational. > I have been overly critical and arguing in bad faith about your writing in the past I think you are just critical without a stated valid reason. Arguing in bad faith seems to be a thing if HN history is the judge. And this is coming from a critical thinker, who is a bit tired of people firing off "human slop" comment…

That’s great for you and Simon but I personally don’t think there will be enough people in the future to use or appreciate these things that can be built and eventually no who will care to build them, and while this “discovery of LLMs phase” might be fun it diminishes the value I used to get from the job, which was writing code, problem solving, discussing and learning with people (having real people and mentors to w…

There are people NOW who use and appreciate it. There are also people who speak for others. I voted you up because you actually put thought into a response, not that the response is a good argument. Some of it is your opinion, and that's yours alone and not for anyone to judge. The rest of it is more focused on others and their actions, before your own. That I can attack, and do so gleefully.

I've been using LLMs for 5 years, sorry, 6 years (sigh). GPT-2 was a dumbass, but I still managed to get it to do primitive function calls - in 2020. I'm not slowing down. I'm increasing speed. It's 2026. Some of you are going to be obsoleted into oblivion if you don't start moving. I estimate I'm at least worth $600K a year now, probably a lot more. And, I don't have a masters. My dad did, and half a Phd, but I digress. I'm not him. Instead of working for the man, I'm doing my own thing, yet again. It is possible for one person to build an empire. I'm going to do it. Not ego. Intent.

The thing you describe here, about it being depressing, is a logical fallacy. I'm not even looking at an AI or Wikipedia. I don't know the proper name for it, but it's a fallacy based on thinking that others bring you value. You clearly have value, you said it yourself and said it well. If anyone sits around and let others define or tell them their value, they are a dumbass.

I've started using "get a horse!" as a shorthand meme for what you are feeling right now, maybe. I'm not here to speak for you or how you manage your emotions. I'm just pointing out the flaws in your thinking. This is what people on horses shouted at people driving the first cars. (the horse thing, although they probably thought driving a car was flawed thinking)

Find your passion and then build it.

Post reply on HN