Top 29 stories from Hacker News. Top 10 include comment highlights. Compiled at 20:10 UTC.
500 points by ilreb · 205 comments
Qwen Studio offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
What HN said:
mynti: To me it feels so weird that people are trying to push these model for online shopping like "here is how this dress/shirt/pants would look on you". But these models will always make the clothes fit your body and show you in flattering light and so on.
weird-eye-issue: The meta keywords in the HTML is very interesting. 100+ references to NSFW topics such as hentai, nudes, etc.
postalcoder: They must have trained on GPT Image 1 outputs. The yellow tint is unmistakable. https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen-Image/i... https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen-Image/i... https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen-Image/i...
hessammehr: Very cool but the Arabic text in the title image is obviously and hopelessly broken, which is oddly not the case when actually using the model. Could it be that the hero image was not generated by Qwen Image 3.0?
471 points by logickkk1 · 373 comments
We’re introducing new Gemini models, including Gemini 3.6 Flash, 3.5 Flash-Lite and 3.5 Flash Cyber.
What HN said:
hmokiguess: Spent half an hour just now benchmarking it against my current 3.5 Flash pipeline excited only see it regressed slightly (0.1% - 0.2% at most, for feature extraction work) Seems like this is mostly a cost play by Google, hoping this doesn't bring 3.
postalcoder: I wonder how big the Pro model is that Google is using behind the scenes to train these smaller ones. Going on baseless speculation, the lack of accompanying pro models with these flash releases either means: 1) the model is too big to be economical, 2) google doesn't have the co...
stonewhite: Google somehow managed to snatch defeat from the jaws of success with their AI products. They literally forced me and my company out of Antigravity by phasing out AI Ultra subscription without any proper product follow-up.
m_w_: It's a bit disheartening to see no comparison to other models here - and I'm not sure this pushes the curve anywhere. 3.6 flash is more expensive than GLM 5.2 - but seemingly worse, although this post is really light (lite?) on details.
232 points by speckx · 205 comments
This case involves Apple’s handling of user-uploaded files hosted in private iCloud storage. Instead of adopting PhotoDNA to scan hosted files for CSAM, Apple created its own proprietary alternative, NeuralHash, which apparently wasn’t as good. So Apple U-turned on its...
What HN said:
giantg2: Maybe my perception is off, but it seems like there's a huge push by the legislature and some people to do anything and everything to prevent CSAM, yet almost nothing seems to be done to prevent CSA. For CSAM, there's all sorts of monitoring, scanning, identify capturing, etc.
amazingamazing: It is crazy people think apple isnt on the side of privacy. Are they perfect? Not even close, but compared to the rest of big tech theyre simply on another level. Apple could easily not do this stuff and it may even be easier to not.
majorchord: IMO "end-to-end encryption" simply isn't possible when the application is run by the same company as the servers the data sits on, is closed source, and can at any time, see the decrypted contents of data it downloads from their servers and do whatever they want with it.
JSR_FDED: The judge called the outcome disturbing, as it leaves victimized children as "collateral damage" of privacy protections. As sad as this is, end to end encryption means no CSAM scanning.
217 points by speckx · 35 comments
What HN said:
bix6: For the coral lovers. There are multiple companies working on reef preservation. They are woefully under resourced. If you can help, please do.
SparkyMcUnicorn: This source appears to have more images and info: https://www.frontiersin.org/journals/marine-science/articles...
F7F7F7: The biodiversity in West Africa in general is completely underrated. I mean, Darwin solidified his trajectory in life a few short days after the Beagle anchored off of Cape Verde (a place he initially detested then fell in love with).
novia: > Zinzindohoué agrees: “We don’t need to wait for others to come to our country to show us what is under our sea. We are the ones who must take responsibility.
151 points by naves · 15 comments
PCjs offers a variety of online machine emulators written in JavaScript. Run DOS, Windows, OS/2 and other vintage PC applications in a web browser on your desktop computer, iPhone, or iPad.
What HN said:
mk_stjames: I just did something neat- I tried out the install of Windows 3.1, which has a version of Visual Basic installed. I created a small program in Visual Basic, with a window and some buttons to press, and then saved the program as an .
glimshe: Those who consider every little JS framework, iPhone and AI iteration "revolutionary", go check the thing labeled "Visicalc (1981)" on this site to learn what a revolution looks like. PS: Nine Princes in Amber was an interesting choice alongside the others!
drivers99: I have my IBM PC set up (to go through a Peter Norton assembly language book again) but I got a file read error for the first time recently on the 20 MB (Seagate ST-225) hard drive and parity errors from the RAM (again).
smusamashah: A somewhat relative list of in-browser VMs https://gist.github.com/SMUsamaShah/6379649c0cbade9abc02b242...
143 points by montecarl · 149 comments
What HN said:
zetanor: I was significantly worried about ChatGPT accepting sponsorships—a necessity in the evolving landscape of AI services—but I've since come to understand that advertisements are not simply burdens, they're opportunities to connect with brands that can fulfill my needs.
sssilver: I always thought their top tier offering to their advertiser customers should be "Inconspicuously, over an extended amount of time, subtly respond to the user in a way that nudges them towards purchasing customer's goods and services, without ever directly mentioning anything.
kamranjon: Pretty bold move to release this at the peak of the Open Models vs Proprietary Models debate. Feels like a really easy decision when it's framed this way.
tux3: Those ads are supposed to be "Clearly labeled" and "Separate from answers". Now, this is the sort of rock-solid commitment to trust and safety that steadily gets worse every year until you go from old Netflix to Poob with ads. But the water's still fine.
135 points by rbanffy · 12 comments
Python 3.15’s Ultra-Low Overhead Interpreter Profiling Mode
What HN said:
abbeyj: I'm not clear on why there is
ip++in the RECORD_INST handler. We've already moved to the next instruction by runningip++at the end of the handler in the normal dispatch table.
NeutralForest: Here are explanation on how to use it from one of the people working on it: https://www.youtube.com/watch?v=f1x4X83CDSA and the chunky docs that go with it in the beta of 3.15: https://docs.python.org/3.15/library/profiling.sampling.html
evomassiny: Could you duplicate
RECORD_INSTbefore eachINSTRUCTION_N, so that:INSTRUCTION_1: // subroutine 1 ip++; goto *dispatch_var[*ip]; INSTRUCTION_2: // subroutine 2 ip++; goto *dispatch_var[*ip];becomes: ``` RECORD_INST_1: //...
haeseong: A
bool profilebranch stays expensive even when it predicts perfectly, because it bloats every opcode handler, and on a computed goto interpreter that extra code messes with the branch predictor history each dispatch site builds up, which is the whole reason dispatch is fast.
129 points by marklit · 100 comments
A free interactive map of 2,435 great castles, fortresses and palaces across 129 countries — each pinned with its photo, founding century and story.
What HN said:
mxfh: If you submit some map, at least don't have the title contradict the content description: "the 2,435 most significant landmarks worldwide, ordered by a fame ranking that blends Wikipedia coverage with real readership.
clickety_clack: This is off by many degrees of magnitude for Ireland at least: https://www.reddit.com/r/dataisbeautiful/comments/1nkvqd7/oc...
tagami: My first dotcom began by listing German castles for sale after the wall fell. https://web.archive.org/web/19970102101502/http://www.re.com...
Bedon292: Cool idea, horrible dataset. Its limited to things in Wikidata rather than using OSM, and it must have an article on Wikipedia in English to count so I am sure it is missing just based on that.
123 points by FriedPickles · 26 comments
Free Ink is an open-source collective building the software, firmware and hardware for e-paper readers — every layer ships in the open for anyone to pick up, extend and make their own.
What HN said:
Varelion: Perfect -- this is exactly the kind of project I was hoping to see gain some traction as of late.
monocularvision: I searched for each of the supported eInk devices and they were all tiny. Anyone know if there is a larger reader (maybe Paperwhite size-ish) that might run this?
imzadi: I got an Xteink X4 a few weeks ago and love it. I've tried out a few different firmwares but haven't settled on one yet. I love the screen and how simple the interface is. It's kind of a pain in the butt getting my Kindle books on there, but worth the effort.
t1234s: Are there any instances from the closed e-book system (kindle, nook, etc..) where you paid for a book and the book was either updated to a revised version or just removed from the device because of the content in the book?
106 points by tchalla · 64 comments
What HN said:
bnfcl: Quote of the main point in the article: In other words, the hard part moves from recall (“How do I write this?”) to judgment (“Does this actually make sense?”) This is very true. But to evaluate if it makes sense, you first need experience writing the code.
FinnLobsien: It’s the same with writing: AI writing is coherent on the surface, but is impossible to edit because it’s built on no real insight. Now the sharpening of a coherent point, challenging one’s assumptions, and editorial decisions of what (not) to include are super important because...
foo12bar: This article will apply for about 3 months until AI advances again and the assumptions made are no longer true. You can't take a rapidly developing field and pretend progress is going to freeze at its current development level so you can decide how to handle it.
austin-cheney: I am looking at the subtitle: The future of software development will belong to those who can think clearly at scale, maintain durable mental models amid rapid change, and integrate machine-generated output into human-directed intent.
AI/ML
Business/Tech
Other
Programming
Society
Stories and comments sourced from Hacker News public API. Not affiliated with Y Combinator or Hacker News.