1. Claude Opus 5.5 (www.anthropic.com)

Claude Opus 5.5 leads in agentic coding and knowledge work, and costs 40% less to run than Opus 5 on typical workloads.

Claude Opus 5.5 offers significant cost savings for coding and knowledge tasks, running 40% cheaper than its predecessor while maintaining high performance. This update could impact budget-conscious teams looking to improve efficiency.

2. GPT-6 Sol and Luna (openai.com)

Meet GPT-6 Sol and Luna, two models that bring frontier intelligence to everyday work with different balances of capability and cost.

GPT-6 Sol and Luna are new AI models designed to enhance productivity by offering varying levels of performance and affordability. These options aim to cater to a wider range of professional needs while maintaining high intelligence.

3. Pentagon: Palantir AI Overreliance Led to Strike Killing 123 Iranian Children (gizmodo.com)

An unreleased Pentagon probe investigated the kill chain that hit a school in Minab.

A Pentagon investigation reveals that overreliance on Palantir’s AI systems contributed to a strike that tragically killed 123 children at a school in Minab. The report raises serious concerns about the implications of using AI in military decision-making.

4. I said no and Apple said yes (dbushell.com)

The one where Apple decides consent doesn’t matter anymore.

Apple’s recent actions raise serious questions about user consent and privacy, suggesting a shift in how the company prioritizes these issues. This commentary highlights the implications of this decision for consumers.

6. Apple has added persistent 'ads' to iOS, and it's driving users crazy (www.techradar.com)

iOS users are complaining that Apple is placing 'ads' for its own products on their iPhones.

iOS users are frustrated as Apple introduces ads for its products directly on their devices, sparking complaints about the intrusive marketing approach. The backlash highlights growing concerns over user experience and privacy.

8. Spymarks, Not Watermarks (brand.io)

Watermarks that spy on users are no mere watermarks

New types of watermarks are being developed that can track user behavior without their knowledge. This raises significant privacy concerns that are worth understanding.

9. Microsoft killed FoxPro in 2007. Anyway, here's FoxPro revived (foxscript.org)

A modern IDE and 64-bit runtime for Visual FoxPro 9 applications: the same language, rebuilt on Electron, React and a Rust virtual machine in WebAssembly.

A new development environment breathes life into Visual FoxPro 9 applications, combining familiar language features with modern technologies like Electron and Rust. This revival aims to provide a 64-bit runtime, enhancing usability for legacy systems.

11. Can gzip be a language model? (nathan.rs)

A while back I wrote about language modeling without neural networks, where I generated Shakespeare with an unbounded n-gram model: no weights, no training, …

The article discusses the potential of using gzip, a data compression tool, for language modeling without relying on neural networks. It highlights an innovative approach to generating text, such as Shakespeare, through an unbounded n-gram model.

12. AI Has No Wisdom and Neither Will You (alexn.org)

Certainly the industry is transforming, however, the people and organizations falling into the trap of no longer reading and writing code only do so at their peril.

The article warns that as the tech industry evolves, losing the ability to read and write code can be detrimental for both individuals and organizations. It emphasizes the importance of maintaining coding skills to navigate the future of technology effectively.

14. SAML: A Fractal of Bad Design (blog.trailofbits.com)

SAML, the XML-based authentication protocol that birthed the SSO industry, is fundamentally flawed due to XML complexity, canonicalization issues, enveloped signatures, and design ossification, making it vulnerable to signature wrapping attacks and parser differentials that persist despite decades of awareness. Organizations should migrate to OpenID Connect (OIDC), which avoids these pitfalls through simpler JSON-based design, detached signatures, and agile evolution.

SAML’s design flaws, including XML complexity and security vulnerabilities, have led experts to recommend a shift to OpenID Connect (OIDC). The article discusses the specific issues with SAML and highlights the advantages of OIDC’s simpler and more secure approach.

15. How Meta's Muse works, revealed by the 6.8 GB filesystem it sent me (mouse.dev)

I asked Muse to archive the files it could see and send them to my Google Drive. It did.

Meta’s Muse successfully archived files and transferred them to Google Drive, showcasing its capabilities in file management. This practical demonstration highlights how the tool can streamline data organization and sharing.

16. Claude Opus 5.5 Intelligence, Performance and Price Analysis (artificialanalysis.ai)

Analysis of Anthropic's Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback) and comparison to other AI models across key metrics including quality, price, performance (tokens per second & time to first token), context window & more.

Get insights into how Anthropic’s Claude Opus 5.5 stacks up against competing AI models based on crucial metrics like performance, pricing, and quality. This analysis provides a detailed comparison to help you understand its strengths and weaknesses.

20. Claude Opus 5.5 (github.com)

Claude support for Apple Foundation Models. Contribute to anthropics/ClaudeForFoundationModels development by creating an account on GitHub.

21. A WordPress vulnerability scored 9.2/10 is present in all versions since 2016 (github.com)

An unauthenticated attacker can make `get_page_template()` page-template resolution include a chosen readable local `.php` file outside the active theme directories. If relevant pre-conditions for ...

A serious vulnerability in WordPress allows unauthenticated attackers to access local PHP files outside of theme directories, posing a significant risk to sites running any version since 2016. It’s crucial for site owners to understand the implications and necessary updates to protect their content.

23. ReBarUEFI: Resizable BAR for almost any UEFI system (github.com)

Resizable BAR for (almost) any UEFI system. Contribute to xCuri0/ReBarUEFI development by creating an account on GitHub.

A new project on GitHub aims to enable Resizable BAR support for a wide range of UEFI systems, potentially enhancing graphics performance for many users. Check out the details to see how it could benefit your setup.

24. Unreal Agent (unreallabs.ai)

We’re sharing Unreal Agent β€” an agent harness that delivers up to 40% cost savings compared to Codex on production workloads and coding/science benchmarks, without any negative performance impact.

25. Show HN: Drop – a rootless Linux sandbox with gVisor support (droprun.sh)

Drop: a rootless Linux sandbox for developers. It isolates coding agents and third-party packages while preserving your work environment.

Drop offers a solution for safely running third-party programs without compromising your main user account. By utilizing gVisor support, it aims to provide a more efficient alternative to traditional containers and VMs, addressing both security and productivity concerns.

26. Line Scan Photos from MUNI Heritage Weekend in San Francisco (daniel.lawrence.lu)

I went to San Francisco for Muni Heritage Weekend 2026!
I stood by the tracks for three hours with my line scan camera, scanning trams.

A photographer shares their experience at Muni Heritage Weekend, capturing trams in San Francisco using a line scan camera. Check out the unique images and insights from this event dedicated to the city’s transit history.

28. Obscura: The first VPN that can't log your activity (obscura.com)

Privacy first VPN. Two-party relay and no logs. No email needed. Block ads, trackers, malware. Available on iOS, macOS, Android, Windows, and Linux.

Obscura offers a VPN service designed with privacy in mind, featuring a two-party relay system that ensures no activity logs are kept. It also includes ad, tracker, and malware blocking, and is compatible with multiple operating systems.

29. 9 Ads per Minute: FIFA Cup 26 – "the price of the beautiful game" (www.bristol.ac.uk)

More than 93,000 visible appearances of brands promoting unhealthy food and drink, alcohol, gambling and speculative products were recorded during coverage of the largest FIFA World Cup in history, new research has found. Unhealthy food and drink brands accounted for more than seven in ten, with 65,722 appearances.

Research reveals that the FIFA World Cup 2026 is inundated with brand ads, particularly for unhealthy food and drink, making up over 70% of the total brand appearances. With more than 93,000 ads noted, this analysis highlights the significant influence of marketing during major sporting events.

30. Show HN: JevBench, a reproducible benchmark for typed decision models (benchmarkheaven.com)

52 Jev-class systems tested on 534 decisions. Jev 1.13.0 leads JevBench v1.3.0 with 74.4; compare open-source, self-hostable and hosted options.

JevBench offers a way to benchmark typed decision models, allowing users to compare the performance of Jev-class models against other projects in terms of accuracy, latency, and cost. With customizable weighting options, it aims to provide a clearer picture of how these models stack up against traditional LLMs.

Article Details

Browse by Date

Type to search HN and engineering blog articles
Press Escape to close  |  ↑↓ to navigate  |  Enter to open