GPT-6 Sol and Luna

1223 OfficialTurkey 2026-09-22 18:41:36 UTC openai.com Llama Share 624

OpenAI has introduced GPT-6 Sol and Luna, more cost-efficient models that offer advanced AI capabilities. These models are designed to be used at different scales, rhythms, and budgets, with improved performance in areas such as coding, computer use, and alignment.

Claude Opus 5.5

1243 km144 2026-09-22 16:41:36 UTC anthropic.com Llama Share 834

Claude Opus 5.5 has been released with significant improvements in performance, safety, and efficiency, surpassing previous models like Opus 5 and Claude Mythos 5.1 on various benchmarks and tests.

'We hacked the FBI:' Hackers say they have data on all FBI employees

434 spenvo 2026-09-22 18:41:36 UTC 404media.co Llama Share 305

A high-profile hacking group called ShinyHunters claims to have breached multiple FBI services, stealing data on all FBI employees and applicants. The stolen data includes personal information such as names, home addresses, phone numbers, and spouse details.

OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005

573 sohkamyung 2026-09-22 14:41:36 UTC cryptocellar.org Llama Share 362

A team led by Carter Leffer used OpenAI's GPT-6 Astra to break a 75-year-old Enigma message, MVUEH, which had resisted solution since 2005. The AI achieved this feat on its own after analyzing the message and using a crib technique, revealing the correct key and plaintext.

ReBarUEFI: Resizable BAR for almost any UEFI system

81 nateb2022 2026-09-22 21:41:36 UTC github.com/xcuri0 Llama Share 24

A UEFI DXE driver enables Resizable BAR on systems without official support, providing performance benefits for Intel Arc GPUs and NVIDIA Turing GPUs. The driver can be built using the provided script and requires modifications to the BIOS or DSDT for some motherboards.

Microsoft killed FoxPro in 2007. Anyway, here's FoxPro revived

197 boredjohnny 2026-09-22 21:41:37 UTC foxscript.org Llama Share 134

Visual FoxPro stopped at version 9 due to its 32-bit foundation, which limits file handling and memory. A new compiler in Rust compiled to WebAssembly allows for a 64-bit process with improved performance.

What California is learning from solar panels built over irrigation canals

130 Jtsummers 2026-09-22 19:41:37 UTC kqed.org Llama Share 177

California is exploring ways to reduce water evaporation from its extensive canal system, which transports clean water to cities and farms. A pilot project called Project Nexus tests solar panels on canals, generating power while limiting evaporation.

SAML: A fractal of bad design

170 aray07 2026-09-22 19:41:37 UTC blog.trailofbits.com Llama Share 95

SAML authentication protocol is being considered for retirement due to its complexity and security issues, with modern alternatives like OpenID Connect (OIDC) gaining popularity. SAML was designed in the early 2000s but has not received necessary updates, leading to a decline in its adoption.

Claude Opus 5.5 Intelligence, Performance and Price Analysis (Max)

240 theanonymousone 2026-09-22 17:41:37 UTC artificialanalysis.ai Llama Share 72

Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback) is a proprietary model that scores 58 on the Artificial Analysis Intelligence Index and has a context window of 1M tokens.

WordPress: Unauthenticated path traversal leading to conditional RCE

160 vntok 2026-09-22 16:41:37 UTC github.com/wordpress Llama Share 84

An unauthenticated attacker can execute arbitrary code on a WordPress site by manipulating the page-template resolution, potentially leading to Remote Code Execution (RCE). The vulnerability affects certain themes and PHP versions.

Pentagon says overreliance on AI contributed to missile strike on Iran school

450 devonnull 2026-09-22 19:41:37 UTC bloomberg.com Full Share 233

Please make sure your browser supports JavaScript and cookies and that you are not blocking them from loading. For more information you can review our and .

The current balance of power in open models

49 gmays 2026-09-22 22:41:37 UTC interconnects.ai Llama Share 15

Chinese AI companies are leading in open-weight models, surpassing American counterparts in capabilities and adoption. The US is losing its lead as the preeminent AI research hub due to China's growing investment and talent.

How did AMD Ryzen get 50% faster in two years?

230 ibobev 2026-09-22 13:41:37 UTC lemire.me Llama Share 86

AMD Ryzen 7 processors from 2022 to 2024 have seen significant improvements, with the Zen 5 processor offering a 47% single-core and 58% multi-core speed boost. The new Zen architecture features increased transistors, wider cores, and improved cache sizes.

Side-stepping the Secretary Problem, unwittingly

53 pvdebbe 2026-09-23 00:41:37 UTC evalapply.org Full Share 7

MUNI Heritage Weekend in San Francisco

166 plun9 2026-09-22 11:41:37 UTC daniel.lawrence.lu Llama Share 43

The user attended Muni Heritage Weekend 2026 in San Francisco, capturing tram photos with a line scan camera. They shared interesting facts about historic trams like Muni No. 1 and Milan tram, despite some photo failures due to technical issues or background distractions.

OpenAI is well positioned to fast-follow Jev

266 JohnBerryman 2026-09-22 15:41:37 UTC arcturus-labs.com Llama Share 196

TypeSafe's Jev has taken the AI world by storm with its fast and accurate classification capabilities, but OpenAI may replicate it soon. TypeSafe's training data and reinforcement learning process are their main moat to protect themselves from being copied.

Unreal Agent

139 trollied 2026-09-22 18:41:37 UTC unreallabs.ai Llama Share 83

Unreal Labs developed Unreal Agent to improve AI agent efficiency by reducing tool-management overhead, achieving up to 40% cost savings compared to Codex and 20% compared to Pi. The harness allows for asynchronous tool calls, enabling agents to respond quickly and be cost-effective without the need for polling or waiting.

Show HN: JevBench, a reproducible benchmark for typed decision models

70 florianstandhar 2026-09-22 19:41:37 UTC benchmarkheaven.com Llama Share 17

JevBench v1.3.0 ranks decision models based on Intelligence, Calibration, Speed, and Cost. The top-ranked model is Jev 1.13.0 with a score of 74.4: Intelligence 85.7%, Calibration 82.7%, Speed 83.3%, and Cost $52 per thousand decisions.

Native apps written in TypeScript and CSS

84 arbayi 2026-09-22 20:41:37 UTC github.com/geastack Llama Share 29

This repository contains app gallery examples for various platforms, including simulator and embedded targets. It uses a gea manifest in package.json to configure compatibility with different platforms.

People hooked on vapes try a new way to quit: cigarettes

118 alephnerd 2026-09-22 12:41:37 UTC bloomberg.com Full Share 115

Please make sure your browser supports JavaScript and cookies and that you are not blocking them from loading. For more information you can review our and .

George Lucas Returns to Earth, Bearing Gifts

87 surprisetalk 2026-09-22 17:41:37 UTC commonedge.org Llama Share 45

The Lucas Museum of Narrative Art, designed by MAD Architects, opened in Los Angeles with a diverse collection of 1,300 objects showcasing humanity's stories and artists. The museum features exhibits on Star Wars, comics, and other narrative art forms in a unique blend of architecture and design.

Markdown in /src

93 perrygeo 2026-09-22 18:41:37 UTC htmx.org Llama Share 47

The author proposes using Markdown as a source of truth for software systems, alongside generated code, to capture intent and behavior. This would involve creating a new directory (/src/md) to store human-authored and curated Markdown files that describe the system's behavior.

The Softness of Metal

3 NaOH 2026-09-23 01:41:37 UTC psyche.co Prefix Share 0

I am grateful no one else is in the room with me. In principle, I’ve no problem with my family seeing me cry: but blinking away hot salty tears while watching Ozzy Osbourne? That seems somehow ludicrous. This is, after all, the self-styled prince of darkness, a trickster-like figure – one part devil, one part joker, one part joke. I thought I knew Ozzy – we all thought we knew him, whether we ...

Obscura: VPN that can't log your activity

82 Flimm 2026-09-22 20:41:37 UTC obscura.com Llama Share 75

Obscura VPN offers a unique, no-logs service using Mullvad's exit servers and QUIC technology for maximum privacy. The app's source code is publicly available on GitHub to verify its claims of complete anonymity.

The JavaScript Midlife Crisis

46 maroun-baydoun 2026-09-22 20:41:37 UTC maroun-baydoun.com Llama Share 29

The JavaScript ecosystem has evolved rapidly, with new tools and languages emerging to improve performance, but this shift may lead to a loss of familiarity among developers. The focus on speed is creating a cycle where newer tools are adopted for their performance benefits, potentially making JavaScript less desirable as the primary choice.

Apple has added persistent 'ads' to iOS, and it's driving users crazy

632 MC995 2026-09-22 14:41:37 UTC techradar.com Llama Share 462

iOS users are upset about Apple inserting ads for its own products into the Settings app, with no clear way to disable them. Many have taken to social media and Reddit to express their frustration over the persistent notifications.

Can gzip be a language model?

375 networked 2026-09-22 06:41:37 UTC nathan.rs Llama Share 147

A compressor like gzip can be used for language modeling without neural networks by scoring continuations based on their compressed length. Gzip's DEFLATE algorithm uses a sliding window to find matches and compresses them, allowing it to predict the next bytes in a text prompt.

The UV index is not the warm sensation of sunlight on bare skin

69 evakhoury 2026-09-22 21:41:37 UTC blog.asciitweezers.com Llama Share 58

Sunburn is triggered by DNA damage from UV light absorption, not the warmth of sunlight. The sun's warmth is a result of visible and infrared light heating molecules in skin, unrelated to DNA damage.

16-bit Intel 8088 chip (c. 1985)

109 rbanffy 2026-09-22 16:41:37 UTC allpoetry.com Llama Share 14

A 40-year-old poem discusses the frustrations of early personal computing, listing incompatible systems and echoing a lifelong disgust with proprietary technology. The poem's switch to a turkey buzzard's mating ritual serves as a commentary on the timeless nature of human experience amidst technological chaos.

Show HN: Training a model to identify AI web content from structure alone

43 jochenmadler 2026-09-22 19:41:37 UTC arxiv.org Llama Share 9

Researchers investigated whether AI-generated text can be identified by analyzing structural signatures. They developed a 214-feature instrument that detected AI posts with high accuracy, even after rewording.