Top 30 stories from Hacker News. Top 10 include comment highlights. Compiled at 20:10 UTC.
865 points by alvis · 479 comments
Opus 5 is a step change improvement for the Opus tier powering long-running agents while delivering improvements in coding and professional work.
What HN said:
postalcoder: I think the most important thing here is not absolute performance. It's that organizations now have access to a Fable-ish model without Fable's 30-day data retention requirement[0].
jjcm: Doing testing with it now, specifically for image->html conversion. Previously Fable was the best at this, followed by Gemini 3.1 pro (a surprising #2, but Google has great vision models).
rb2e: https://www.anthropic.com/news/claude-opus-5 - A blog post for those not wanting to go through a 190ish page pdf
paxys: Looking at all these releases it’s not a surprise that model routing is the fastest growing segment in AI right now. There are 10+ LLM companies, each with dozens of models of different modalities, each model with multiple size variants, then different “thinking” levels, then age...
639 points by peykar · 360 comments
I’m feeling it right now. I had to set a timer for 15 minute on my computer and block all distractions to write this. If I didn’t force myself to focus I would easily get distracted by something after few minutes.
What HN said:
projektfu: Hallowell and Ratey, who write about ADHD, suggested a new entity for the current world: VAST, the variable attention stimulus trait. It is like ADHD but it is culturally induced and not presumed to be due to an innate deficiency in executive function.
lardosaurusrex: People often question me how I manage to focus for more than a few minutes and I'm always confused because back in the day I was constantly informed I didn't have any focus and needed to be treated (adhd).
luskira: I've tried multiple things to regain focus, the only ones that worked: - blocking social media everyday til 2pm(i like working from 5-1 everyday), i use freedom app for that - having a "distraction" tablet with 2fa on the social media I use, so I can only use distracting things t...
lrvick: I gave up my smartphone 6 years ago because I concluded the conveniences it offered came at cost of too much of my attention, freedom, and privacy. My ability to focus was just gone, and this magic device was not making me happy or productive anymore.
535 points by ThouYS · 125 comments
FLUX 3, our new multimodal frontier model, jointly learns from images, video, and audio to build one representation of the world. Now available in Early Access.
What HN said:
user43928: I hope the open-weight versions will be SOTA. > Over the next few weeks and months, we will make the following capabilities available > Open-weight access to a multimodal backbone, for content creation (video, audio and image) and action prediction.
thisisauserid: - Showed close to zero examples of people. - Frivolous use of the term World Model. - Claims 20 seconds of video, shows only jumpcuts. Coming soon!
jdthedisciple: It's incredible how negative and dismissive the comments in here are while here I am thinking the model actually looks impressively capable. But then again I heard the downers have always been the first to leave their dung comments here so let's see...
forgotusername6: Is anyone feeding models touch data? It seems the main thing we want the robots to do is touch things, but we are just feeding them audio/video/images. The model has to learn how to touch things despite never having touched anything before.
414 points by hhh · 153 comments
i dissected some firmware for a Hanwha Wisenet XNP-9300RW and found that it had github admin tokens in it, also miltech is weird
What HN said:
badatnames: Worthy thread to ask: is there such a thing as a white label IP camera (or similar) with a supported open firmware? Not asking for open source, but something close to plug and play that nonetheless has a way of stripping the rootfs as desired for bespoke use in a manufacturer-sup...
grommz: The US Department of War IP adresses baked into the firmware is the bigger story here. Note to self: never buy a Korean security product.
dev_l1x_be: Not surprised, many of these vendors are doing crazy things, insane defaults, broken security, hardcoded values. Security is not a priority, I get that, but at the very least some baseline check would be nice (no hardcoded credentials for starting)
tehlike: A rule of thumb, put your cameras on a separate VLAN and never give that vlan internet access. Least you can do.
372 points by sohkamyung · 106 comments
On the first attempt, reaching orbit, I never thought it was possible."
What HN said:
mayama: Regarding Indian private space sector, In addition to skyroot solid rocket, following are noteworthy Agnikul is planning to launch semi-cryo kerosene rocket pretty soon. Their engine is 3d printed and use electric pump fed 25kN small engine.
itissid: There is just so much that can go wrong with rocket launches if you do cursory reading about these things. A few random examples: The lowest stage cant just be "turned down" or modulated because they work by burning an inner solid rubber lining that's like an annular cylinder.
zbentley: And with $160MM in funding at a $1.1B valuation? I don't know about their debt, but to get to LEO in 8 years on that little money is extremely impressive.
sidcool: Only the third country to achieve this. Kudos. There's another startup trying reusable rockets.
301 points by surprisetalk · 263 comments
24usqpep0ejc5w6hod3dulxwciwp0djs6c6ufp96av3t4whuxovj72wfkdjxu82yacb7430qjm8adbd5ezlt4592dq4zrvadcn9j9n-0btgdzpiojfzno16-fnsnu7xd" /> <link rel="preconnect" href=" /> Em dashes are fucking amazing - by Sasha Putilin <meta data-rh="true" name="theme-co...
What HN said:
douglee650: Author failed to explain the dash universe and their semantic assignments. - em dash: adjoining thought (Hey, so — ) - en dash: expresses a range (1–100) - dash: compounds word-parts (Sarbanes-Oxley) I'd also like to see more concise philosophy around em v.
WillAdams: One of the more memorable and influential library finds for me was an article in Advertising Age which enumerated all of the special characters which should be set so as to accurately communicate, so in addition to en and em dashes it noted: - − (minus symbol) for use in math e...
usui: You might think em dashes are amazing—but there is something three times better⸻the "three-em dash" (U+2E3B). Why this thing exists, I do not know. Unicode people seem to like making bad decisions, like Han unification and emoji sprawl.
koolala: Test: 'There is no greater joy in life than typing a sentence, realising it needs a clarification, a caveat, an addendum, but in the same sentence, not a new one, and, guess what, the comma is right there for you.
292 points by kensai · 47 comments
FLUX 3 is running robots. Built with mimic and deployed at Audi, FLUX-mimic shows content creation and physical AI share one foundation: a model that understands the world.
What HN said:
vessenes: Really interesting. Upshot: a well trained multimodal video generation model has a world representation model trained inside it. They’ve done some work lifting this world model out and deploying it to robots, where it seems to work well.
GiffertonThe3rd: The video at around 3.30min, where the robot arm took 3 attempts to reseat the window trim, was quite unnerving - I have not seen such resolving before. Is it new or am I way out of the loop?
altern8: This is awesome. What's sad to me is that we have all this awesome technology, but movies are worse than ever. I usually watch movies from decades ago just to find something decent and it's amazing how with goofy-looking puppets the storytelling was 1000 times better.
flufluflufluffy: Ok the phrasing here, it’s, it’s just - > However, compared to more specialized approaches for representation learning they produce less disentangled representations, which puts a ceiling on their usefulness for tasks that require world understanding.
274 points by rwmj · 127 comments
If OpenAI loudly proclaims how dangerous AI is, investors will hear how powerful it is. And who benefits from that?
What HN said:
Zsfe510asG: Finally mainstream news understands. The unfiltered version: 1) The AI failed to solve ExploitGym problems. 2) The OpenAI sandbox is such a horrible hack that the AI managed to escape using standard and well documented script kiddie methods.
ACCount37: By now, I'm pretty confident that some people would keep screeching "it's just a marketing stunt, AI capabilities and AI risks aren't real, they're just doing this to prop up their stocks" even if they find a Cyberdyne Systems T-800 armed with a shotgun breaking down their front...
dwoosley: There seems to be three popular ways to view this incident. 1. The way OpenAI seems to want: Their latest LLM is too powerful and can’t be contained without them building in guidelines to the model. 2.
bluGill: I don't care how it happened someone should be arrested for illegal intrusion. Agents don't work on their own, someone is responsible. If nobody else the CEO for allowing something unsupervised.
249 points by louiereederson · 148 comments
Chinese open-weight models are gaining steam against leading offerings from American companies. OpenAI and Anthropic did not sign the letter.
What HN said:
dang: From yesterday: Startup founders urge U.S. government not to shut off Chinese open weight AI - https://news.ycombinator.com/item?id=49023016 - July 2026 (841 comments) Also related: OpenAI and Anthropic unite against open-weight AI risks to their bottom line - https://news.
novaleaf: Some irony: I subscribe to Claude and Codex (20x plans), and now Kimi. Why Kimi? because K3 is the only frontier model I can have a serious conversation with about my product's security. (I did apply for OpenAi's Cyber Pilot but got no response)
Linserin: Just saw on Microsoft https://www.microsoft.com/en-us/corporate-responsibility/top...
GodelNumbering: I posted earlier on reddit that at this point, the closed source lobby wanting to ban open weights is effectively outgunned (musk also publicly supported this). The oddly reminds me of back in the day when SOPA had caused a similar stir (https://en.wikipedia.
207 points by m0do1 · 32 comments
Hope eventually will came out!!
What HN said:
smallstepforman: X512 is an amazing treasure in the Haiku community. Other than nVidia hardware accelerated drivers, he ported Haiku to RiscV, got sound through HDMI and DisplayPort working, he got AMD Vulkan drivers for Southern Islands chipsets also working, made break throughs in dozens of oth...
LorenDB: Looks like this may have been based on the nillerusr Source engine, which is based on a 2020 leak of the Source sources. nillerusr's engine has also been used to port some Valve games to Android.
m0do1: As a long time BeOS fan (and user back in the day) I really enjoy reading about progress on Haiku. Similarly cool as the Half-Life 2 port is the work on the ARM platform. With early support for M1 Macs (https://discuss.haiku-os.org/t/my-haiku-arm64-progress/19044...
tracker1: Nifty to see... I always liked BeOS/Haiku from a tech perspective, but really don't care nearly as much for the UX or ergonomics... I really like a lot of the Windows 10 start menu and taskbar, but I do hate the MS search results/ads,etc so I switched to Linux for non-work and mo...
AI/ML
Other
Programming
Security
Stories and comments sourced from Hacker News public API. Not affiliated with Y Combinator or Hacker News.