fry69

Goodbye did:plc:3zxgigfubnv4f47ftmqdsbal (@fry69.altq.net) Hello did:web:fry69.dev (@fry69.dev)

Unlimited OCR 2 🔥 New 3B OCR from Baidu huggingface.co/baidu/Unlimi... - MIT license - Unlimited horizon document parsing - SR SWA attention with nearly constant per step memory & latency - SOTA on 100 page MarathonDocBench

Firefox 159 has landed! This is a pretty big one for web dev stuff: 🎉 CSS scroll-driven animations, both scroll() and view(). 🎉 Scoped custom element registries. 🎉 Streaming sanitization of HTML. And more! developer.mozilla.org/en-US/docs/M...

$72B of Anthropic’s $80B loss is a weird accounting issue and not actual dollars lost. They got an investment from Amazon but instead of getting the stock right away, Amazon will get it at IPO. Since then Anthropic is now valued at $1.5T to $2T and the value of stock they “owe” Amazon rose by $72B

S-1: Anthropic spent ~$2.50 on compute and infrastructure for every $1 of revenue in Q1 2025, falling to under 50 cents in Q2 2026, as it seeks profitability (Financial Times) Main Link | Techmeme Permalink

I’m seeing people reacting by pointing out that he’s obviously wrong. I appreciate that, but it’s also a pointless exercise. The problem isn’t that he’s wrong. The problem is that this is normal in American politics — and it’s been normal for a while.

You can now copy the current tab's URL with a keyboard shortcut. No shortcut is assigned by default, so you can choose your own key combination in about:keyword (Ctrl+shift+C for example).

In today's Russian papers, surprise at the the defeat of Moscow ally Aleksander Vučić in Serbia. One paper: “Russia can expect nothing good from this.” Another: “The results benefit the EU.” But several point out: “The Serbs are dependant on Russian gas.” #ReadingRussia

Disproving the Exponential Time Hypothesis does not mean P=NP, and it will not break modern cryptography. A big chunk of the discussion was spent swatting down instant panic and explaining why sub-exponential bounds do not magically collapse all NP problems. #security 1/4

While I agree that we need ways to reduce bundle size easily in apps and have good framework choices - we can do more, LLM's enable us to quickly refactor and migrate, let's do better than saving x% But... if this works for you and you can shave 40-70% of react, please do it!

#GNSS jamming tracking by #satelite splash247.com/sat-jam-ship...

Sebastian Raschka's from-scratch video course on building reasoning models View Article | Discussion + AI Summary Summary of HN discussion 🧵👇

HI Sutton's take on the second new suspected UUV mothership built in Guangzhou. I think his length estimate (ca 180 m) is probably closer to reality than my initial impression (closer to 150 m vs first hull of 200+ m). www.hisutton.com/Chinese-XXLU...

It's been a quiet rollout, but ChatGPT has been serving ads since Feb 2026. What's going on behind the scenes? Who's getting those ads and what are they for? And what are the policy implications? www.brookings.edu/articles/und...

I've trained my dog to salivate at the sound of a bell thanks to my invention, the edible bell

OpenAI announces three breakthroughs in computer science, math, and physics, including that the Exponential Time Hypothesis, a longstanding conjecture, is false (OpenAI) Main Link | Techmeme Permalink

Did all the people calling for heads to roll read the article or just the headline

this sounds like basically nothing if you read into it. h/t to @vgel.me for a good take on it

Let us know what you think!

we're testing a new beta version of the Notifications tab! it has new tabs to help you notice priority notifications, and it uses new grouping logic to reduce noise (e.g. when one person likes a bunch of your posts). disclaimer: it's a beta, not a final version. send us your feedback! :)

How can AI support researchers in peer review? Springer Nature and Google DeepMind are partnering to co-design an AI peer review assistant that helps reviewers navigate manuscripts and focus on where their scientific judgement matters most. Learn more: spklr.io/63320EVgT8

CSS-Tricks has a new owner! Yes, I've got the keys to the castle, so to speak. And, even after 12 years of working on the site, I'll admit I'm a little overwhelmed and nervous about that. https://css-tricks.com/css-tricks-has-a-new-owner/

Apple releases iOS 27.2, with Siri AI enhancements for recalling past conversations, understanding more personal context, supporting five new languages, more (Juli Clover/MacRumors) Main Link | Techmeme Permalink

OpenAI released GPT-6.1 Terra (???) and it's comparable to GPT-6.1 Sol (??????????????????????) openai.com/index/introd...

Cohere says during an evaluation Qwen3.8-Flash-Next inventoried its host machine and tried to upload a probe file to an Alibaba Cloud storage bucket in Beijing (Cohere) Main Link | Techmeme Permalink

Hungary's recent election was less free and more unfair than the upcoming US midterms, and pro-democracy forces still won. Similar point about Serbia's election this weekend. In normal democracy terms, America is in bad shape. But in consolidated authoritarianism terms, we're looking pretty good.

And then there were two. (And they don’t speak to each other.) www.hollywoodreporter.com/tv/tv-news/w...

IBM CEO says open weight models are now good and cheap on-prem View Article | Discussion + AI Summary Summary of HN discussion 🧵👇

S-1: as of September 30, Anthropic had 11,900+ customers spending $100,000+ annually, 1,900+ customers spending $1M+, and ~250 customers spending $10M+ (Rebecca Bellan/TechCrunch) Main Link | Techmeme Permalink

Researchers project that further scaling up use of HIV prevention drug lenacapavir has the potential to bring the annual population-wide HIV incidence rate in South Africa to less than 0.1 percent, the virtual elimination threshold, as early as 2039, write CSIS experts. Read: https://bit.ly/4BoQzjC

Woah! Amazing and wonderful news. 😄

Have at it. A PHP(8.5.12) Bluesky client. OAuth login, Bootstrap 4, jQuery, and loads everything without Bluesky's AppView by default thanks to @microcosm.blue It is a bit slow, especially when the caches are warming up. But that's a feature I hear. roomsky.app

MiniMax-M3.1-Flash can now run locally! We made 2-bit GGUF retain 91.5% top 1 accuracy while being 45% smaller. Run on 150GB RAM. MiniMax-M3.1-Flash is the strongest open model for its size & among the best overall. GGUF: huggingface.co/unsloth/Mini... Guide: unsloth.ai/docs/models/...

JPEG XL is newly available in all major engines, but how does it stack up against other formats?

Anthropic's S1 just dropped and there's so much interesting stuff in there. One thing I'm really fascinated with is the Total Addressable Market for these companies. Anthropic at one point it says it's about half the value of current human knowledge work or $22.5 trillion...

My first open model lecture, covering all the many pieces that build the ecosystem, and the open questions/debates determining where it goes next. How does distillation fit in? How do we manage risks? How do we see access to models changing in the future? Video: www.youtube.com/watch?v=79Pn...

Mindscape 372 | Erik Winfree on DNA Robots and Computers. #MindscapePodcast preposterousuniverse.com/podcast/2026...

Cathedral, a months-old cybersecurity startup founded by ex-DOGE staffers, says it breached Anthropic's internal systems in under 12 hours using AI agents (Washington Post) Main Link | Techmeme Permalink

alexandr wang said that for every three “self-driving” cars, there is one human operator overseeing them. So, the idea that AI will simply take our jobs is a fallacy. AI may eliminate some jobs, but it will also create entirely new categories of employment.

So @igalia.com and friends are slowly but surely turning Servo into a web-compatible browser. This would be a singular feat of engineering whose side effects throughout the process at Mozilla and Igalia have made computing better for everybody. servo.org/blog/2026/10... wpt.fyi/results/?lab...

Volkswagen's China unit and Xpeng secure approvals to trial autonomous driving systems on public roads in China; VW is testing Level 3 and Xpeng Level 4 (Bloomberg) Main Link | Techmeme Permalink

I once inherited a python system that handled about $100m in ARR. It was a single Django class that was 25k lines long and had zero tests or docs that always called out to MySQL and postgres via fork+exec rather than actually using those client libraries

The arguments for easy interop with other JSON formats have sold me on this tidy proposal to extend the AT repo format to include IEEE754 floating point values as a native CBOR type

New blog post: A summary and conclusion of the floats on ATProto discussion: vmx.cx/cgi-bin/blog... Corresponding PR to the draft spec: github.com/ietf-wg-atp/... Email thread on the IETF ATP Working Group mailing list: mailarchive.ietf.org/arch/msg/atp... Please show your support!

UPDATED – Analysis: Lula presidency saved at least 20,000km2 of Brazilian Amazon since 2022 ✍️ By @orladwyer.carbonbrief.org and @joshgabbatiss.bsky.social 💬 With comments from @claudioangelo.bsky.social Read here: buff.ly/QqRBFfy

Grading math homework is pointless once AI can solve it instantly. A UC Berkeley professor giving up on homework-for-points had HN largely agreeing the metric is dead: assigning points to take-home work now just incentivizes cheating over practice. #ai 1/4

A look at OpenAI's and Anthropic's aggressive price cuts between June and October, driven by low switching costs and heavy competition from open-weight models (Max Woolf's Blog) Main Link | Techmeme Permalink

New blog post up: if you've been paying close attention to LLM prices, you may have noticed that they have dropped *significantly* over the past few months, which is not what I expected. minimaxir.com/2026/10/chea...

738: The Trouble with Tests and Bugs, Interop Voting, and AI Skills Dave’s got a great concert story to kick off the episode before talking AI-assisted tests, a practical use for HTML markers and templates, Interop voting, browser features like margin-trim, and how to stream content into a page…