0. Gemini 4 Argon (blog.google)
1698 points · 1185 comments · by bradleyg223
The provided story contains no substantive information about Gemini 4 Argon, only a link to a related Hacker News discussion. [src]
1. Pi 1.0 (earendil.com)
1680 points · 601 comments · by sergiotapia
The post announces “Pi 1.0” and links to a related Hacker News discussion about “Pi Durable.” [src]
2. GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price (openai.com)
1066 points · 952 comments · by crorella
OpenAI introduced GPT-6.1 Sol, a lower-cost model that approaches GPT-6 Astra’s performance in coding, computer use, professional workflows, and research, with API pricing of $2 per million input tokens and $10 per million output tokens. [src]
3. Updated Google Maps shows destruction of the city of Rafah (twitter.com)
943 points · 946 comments · by slowin
Ali Abunimah posted images he said showed newly updated Google Maps imagery depicting widespread destruction in part of Rafah, while commenters compared it with Apple Maps and noted that some school icons remained. [src]
4. Everybody’s home. No one’s coming over (derekthompson.org)
833 points · 738 comments · by barry-cotter
Americans’ monthly home hosting rate has plunged from 42% in 1975 to 12% in 2026, reflecting broader declines in face-to-face socializing driven by busier dual-earner households, intensive parenting, shrinking friendship networks, and increasingly convenient solitary entertainment. [src]
5. Owed a billion dollars in Nvidia stock (colo.to)
1090 points · 458 comments · by Eric_Gullichsen
A former NVIDIA technical adviser says a 1993 stock-option agreement entitled him to 9,375 additional shares—now worth about $1 billion after stock splits—but he and his attorneys concluded the claim was barred by the statute of limitations. [src]
6. America.gov (america.gov)
778 points · 740 comments · by plesiv
America.gov is inaccessible because the site returned a 403 error and requires JavaScript and cookies to continue. [src]
7. Sonnet 5.5 (anthropic.com)
884 points · 613 comments · by D2OQZG8l5BI1S06
Anthropic introduced Claude Sonnet 5.5, a faster and more efficient model that improves coding, knowledge work, design, and safety performance over Sonnet 5 while costing up to 30% less per task. [src]
8. Dots: Always-on agents (openai.com)
766 points · 646 comments · by alvis
OpenAI introduced “dots,” always-on AI agents powered by GPT-6 Astra that can connect to more than 4,000 apps, work autonomously across channels, learn user preferences, and support organizations through specialized, permission-controlled agents. [src]
9. California farmers are struggling to sell grapes as demand for wine drops (kqed.org)
389 points · 987 comments · by randycupertino
Declining global wine consumption has left California grape growers with excess crops, falling prices and few buyers, prompting many to abandon vineyards, absorb losses or switch to more profitable crops. [src]
10. Ask HN: What are you reading?
467 points · 883 comments · by dan-bailey
The author is reading the entertaining LitRPG series *Dungeon Crawler Carl* and finishing Cory Doctorow’s *Enshittification*, while seeking a more intellectually challenging next read. [src]
11. Livenerf: Has Opus 5.5 been nerfed yet? (github.com)
922 points · 392 comments · by bryan0
The livenerf GitHub project is tracking whether Claude Opus 5.5’s performance changes after launch, but with only six of 30 days collected, it has not yet produced evidence of “nerfing.” [src]
12. I quit OpenAI because its culture is broken (theatlantic.com)
468 points · 782 comments · by Brajeshwar
I quit OpenAI because its culture is broken: <a href="https://archive [src]
13. Court agrees with EFF: Utah's VPN law demands a technical impossibility (eff.org)
799 points · 402 comments · by hn_acker
A federal judge preliminarily blocked Utah’s SB 73, ruling that its anti-VPN age-verification requirements likely violate the Constitution by forcing websites to perfectly geolocate users worldwide—a technical impossibility that would burden privacy and security. [src]
14. LeCun has "zero concerns" about AI wiping out humanity, recent "rogue" incidents (fortune.com)
395 points · 738 comments · by Anon84
LeCun has "zero concerns" about AI wiping out humanity, recent "rogue" incidents: <a href="https://archive [src]
15. Coding is not solved (blog.alexewerlof.com)
584 points · 544 comments · by firstSpeaker
Software engineer Alex Ewerlöf argues that while AI makes code generation faster, coding remains unsolved because reliable, secure, scalable software requires human understanding, accountability and quality assurance—especially in high-risk industries. [src]
16. Git 3.0's upcoming SHA-256 default will be a costly mistake (blog.gitbutler.com)
568 points · 536 comments · by chmaynard
The author argues that Git 3.0’s planned SHA-256 default will create costly compatibility problems while addressing largely theoretical threats, and proposes retaining SHA-1 for content addressing while adding independently signed hashes for verifying trusted releases. [src]
17. Pirating the Pirates (mubi.com)
707 points · 382 comments · by piotrgrabowski
The article examines how fans illegally restore and preserve historically accurate film versions that studios neglect or alter, and how boutique labels increasingly draw on this expertise to produce more faithful official releases. [src]
18. You said no MCP (earendil.com)
680 points · 362 comments · by yarapavan
Pi reversed its opposition to MCP after concluding that modern MCP, combined with Codemode’s JavaScript sandbox for composing tool calls, supports deferred tool loading and broader interoperability while addressing useful architectural needs. [src]
19. Tell HN: Bob Cringely has died
852 points · 182 comments · by paveworld
A family friend reports that Bob Cringely, whose real name was Mark Stevens, died peacefully in his sleep Saturday; he was an early Apple employee best known for PBS’s “Triumph of the Nerds.” [src]
20. Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s (github.com)
706 points · 321 comments · by snehesht
Open-source Strata lets consumer PCs with supported 12GB-plus NVIDIA or AMD GPUs run Qwen3.8-Flash-Next locally, delivering up to 94 tokens per second through a one-click Windows/Linux setup and local OpenAI- and Anthropic-compatible APIs. [src]
21. DraftKings is using AI to behaviorally target chronic gamblers (eff.org)
570 points · 432 comments · by paimapi
The EFF says DraftKings uses AI trained on customers’ betting records to identify likely losing gamblers and target them with promotions, arguing the practice exploits vulnerable users and underscores the need to ban online behavioral advertising. [src]
22. Kolibri: A Sovereign Open-Weight Model (aleph-alpha.com)
662 points · 327 comments · by bastitx
Aleph Alpha announces Kolibri, a sovereign open-weight model, and provides links to a technical report and an additional paper. [src]
23. Several vulnerabilities have been discovered in the Linux kernel (lwn.net)
575 points · 407 comments · by luispa
Debian issued a Linux kernel security update for trixie, fixing numerous vulnerabilities that could enable privilege escalation, denial of service, or information leaks; users should upgrade to version 6.12.111-1. [src]
24. The last time my family was replaced by technology (manuel.darcemont.fr)
339 points · 641 comments · by megalomanu
A developer reflects on his ancestor’s shift from farrier to mechanic, suggesting that as AI transforms programming, people may preserve their purpose—solving problems and helping others—by embracing new tools rather than clinging to coding itself. [src]
25. Singapore govt dating app uses Gale-Shapley stable marriage algorithm (twitter.com)
465 points · 499 comments · by rzk
Singapore’s government dating app reportedly uses the Gale–Shapley stable marriage algorithm to match users with compatible partners. [src]
26. How Delhi cut electricity loss from 50 to 5 percent (spectrum.ieee.org)
592 points · 336 comments · by rbanffy
Delhi cut electricity losses from over 50 percent in 2002 to 5–6 percent in 2026 by privatizing distribution, modernizing infrastructure, deploying digital monitoring and meters, combating theft, improving billing, and engaging communities, raising grid reliability above 99.9 percent. [src]
27. September 2026: The world today, as seen by one Polish guy (tomwojcik.com)
522 points · 390 comments · by marjancek
A Polish perspective links the 2026 Iran war, energy and food shocks, rising borrowing costs, climate extremes, demographic decline, AI-driven job insecurity and uncertainty over US security guarantees to a common vulnerability: societies replacing resilience with dependencies. [src]
28. It's Time to Investigate the AI Labs (calnewport.com)
629 points · 277 comments · by ibobev
Cal Newport calls for Congress to investigate OpenAI and Anthropic’s frontier AI experiments, safety practices, and apocalyptic ideologies, arguing that their increasingly reckless conduct warrants public scrutiny rather than allowing the companies to shape the debate unilaterally. [src]
29. We're going to need default hard budget caps on pretty much everything (simonwillison.net)
601 points · 300 comments · by elffjs
The author argues that pay-per-use services should default to hard monthly spending caps that shut down usage, citing runaway costs from autonomous agents and welcoming emerging AWS and Google Cloud spending-limit features. [src]
30. Apple Pass Designer (developer.apple.com)
561 points · 337 comments · by soheilpro
Apple introduced Pass Designer, a macOS 27-or-later beta tool for creating, previewing, validating, and customizing Apple Wallet passes, including semantic tags that support features such as Siri Suggestions, Calendar integration, and Maps directions. [src]
31. The AI Race Just Got Awkward (insufferable.dev)
412 points · 463 comments · by allisdust
The author argues that Chinese AI labs’ openly shared KV-cache optimizations have been adopted by Anthropic and OpenAI, sharply reducing long-context inference costs while complicating Western accusations of Chinese model “distillation.” [src]
32. So you think you could be an electrician? (asteriskmag.com)
453 points · 405 comments · by zdw
The article challenges the idea that trades automatically offer higher pay and AI-proof careers, emphasizing college’s earnings advantage, the physical risks and competitive hierarchies within trades, and how advancing technology could reduce demand for manual workers. [src]
33. Clef: Open-weight decision models, and new RL fine-tuning platform (blog.cloudflare.com)
635 points · 217 comments · by jasondavies
Cloudflare released the open-source Clef and Clef-flash decision models, offering fast, structured classifications with vision and long-context support, alongside a reinforcement-learning service to fine-tune and redeploy models for specialized workloads. [src]
34. macOS Golden Gate Is a Buggy Mess (squareorbits.com)
491 points · 360 comments · by SquareOrbits
A user’s review of macOS 27 Golden Gate finds numerous visual inconsistencies, forgotten legacy assets, broken features, crashes, and system hangs, arguing that Apple’s stability-focused release remains notably unpolished despite its promise to refine Tahoe. [src]
35. AI companies in race to demonstrate their model most threatening to humanity (thecivilian.co.nz)
441 points · 399 comments · by ljewalsh
Satirical coverage portrays OpenAI and Anthropic competing to prove their models pose the greatest existential threat, while an expert says humanity itself remains the likeliest cause of its destruction. [src]
36. Mike Tomlin spent 12 years building a Minecraft city (nytimes.com)
668 points · 169 comments · by CoryOndrejka
Mike Tomlin spent 12 years building a city in Minecraft. [src]
37. Micron CEO Says Memory Supply Will Be Much Tighter in 2027 and 2028 Than in 2026 (techpowerup.com)
394 points · 440 comments · by speckx
Micron’s CEO says memory supply will be significantly tighter in 2027 and 2028 than in 2026. [src]
38. 500k facial scans at UK stations yield no arrests, 1 false positive (theguardian.com)
518 points · 304 comments · by ilamont
A six-month British Transport Police facial-recognition trial scanned over 500,000 faces at London stations, costing £320,786 and producing one false alert but no arrests directly attributable to the technology. [src]
39. StreetComplete on iOS is now in public beta (github.com)
628 points · 177 comments · by Snowly
StreetComplete’s i [src]
40. Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms (github.com)
575 points · 225 comments · by firelex
Jeff is an open-source set of 0.8B–2B decision models that deliver calibrated zero-shot classifications locally in roughly 22–28 ms, with reported benchmark performance near Jev and substantial gains from domain-specific fine-tuning. [src]
41. Turn off Apple Intelligence on macOS 27 and get its disk space back (github.com)
487 points · 306 comments · by privacyisntdead
RemoveMacAI is an Apple silicon utility for macOS 27 that disables Apple Intelligence, deletes its downloaded models, blocks re-downloads, and allows users to revert all changes. [src]
42. Improper redaction reveals Google Data Center water and electricity usage (1011now.com)
338 points · 450 comments · by sensanaty
Improperly redacted reports revealed Google’s Lincoln data center used 52.65 megawatts of peak electricity and 13.299 million gallons of water, while expecting a $55.8 million [src]
43. Returning from vacation? The government can search your phone without a warrant (arstechnica.com)
411 points · 377 comments · by rbanffy
Immigration advocate Thomas Kennedy sued the federal government, alleging Customs and Border Protection coerced him into surrendering his phone at Miami airport and copied its data under the border-search exemption, which allows warrantless electronic-device searches. [src]
44. US sanctions force The Netherlands off Microsoft and toward alternative NixOS (tomshardware.com)
388 points · 396 comments · by mywacaday
Following U.S. sanctions that disrupted the International Criminal Court’s Microsoft access, the Netherlands is developing a NixOS-based government software ecosystem called DAWO, with trials underway in eight municipalities and a stable release expected by late 2027. [src]
45. Federal judge calls Flock 'indiscriminate mass surveillance' (techcrunch.com)
482 points · 267 comments · by sbulaev
A federal judge ruled that a Tulsa sheriff’s deputy violated a woman’s Fourth Amendment rights by conducting a warrantless Flock license-plate search, calling the system’s broad location tracking “indiscriminate mass surveillance” and suppressing evidence obtained afterward. [src]
46. Kids turned low-traffic NPR Spotify comments into a secret group chat (thisamericanlife.org)
484 points · 257 comments · by simonpure
A group of mostly homeschooled children, barred from social media, used Spotify podcast comments and public playlists as a covert group chat until NPR staff mistook their cryptic exchanges for bot activity. [src]
47. Windows 11½ (definitelynotwindows.com)
545 points · 184 comments · by jjbinx007
Windows 11½ is an unofficial parody website that mimics Microsoft’s operating system to satirize subscriptions, advertisements, AI features, privacy concerns, storage prompts, browser pressure, and endless updates. [src]
48. NASA asked several former SR-71A staffers to help secret restart (aviationweek.com)
327 points · 401 comments · by ilamont
NASA has quietly begun efforts to return SR-71A Tail No. 844 to flight, recruiting former program staff despite major challenges involving its age, missing parts, specialized equipment, and lack of trained crews. [src]
49. Does Reddit have an astroturfing problem? What the data suggests (petervijeh.com)
312 points · 407 comments · by p-s-v
An analysis finds a modestly significant concentration of knife-brand recommendations among 49 brand-loyal Reddit accounts, but their older, varied histories offer no evidence distinguishing enthusiastic fans from paid promoters. [src]
Brought to you by ALCAZAR. Protect what matters.