Google releases Gemma 4 open models

510 jeffmcjunkin 2026-04-02 16:32:23 UTC deepmind.google Full Share 141

Build autonomous agents that plan, navigate apps, and complete tasks on your behalf, with native support for function calling. These models were evaluated against a large collection of datasets and metrics to cover different aspects of text generation. See additional benchmarks in model card.

Qwen3.6-Plus: Towards real world agents

269 pretext 2026-04-02 14:32:23 UTC qwen.ai Llama Share 91

Qwen3.6-Plus is a significant upgrade to the Qwen series, offering enhanced agentic coding capabilities, improved multimodal reasoning, and a more stable and reliable foundation for developers. The model is now generally available through the Alibaba Cloud Model Studio API and supports various coding assistants, including OpenClaw, Claude Code, and Qwen Code, to streamline development ...

LinkedIn is illegally searching your computer

1206 digitalWestie 2026-04-02 13:32:23 UTC browsergate.eu Llama Share 560

LinkedIn scans users' computers for installed software without consent, collecting data on job search activity, religious beliefs, and competitor products. BrowserGate is a campaign to expose this corporate espionage and data breach scandal, seeking funds and public support to hold LinkedIn accountable.

Lemonade by AMD: a fast and open source local LLM server using GPU and NPU

313 AbuAssar 2026-04-02 11:32:23 UTC lemonade-server.ai Llama Share 80

Using --no-mmap and increasing context size can speed up load times. Lemonade offers a unified API with various features like chat and image gen.

OpenAI Acquires TBPN

38 surprisetalk 2026-04-02 17:32:23 UTC openai.com Llama Share 32

OpenAI acquired TBPN, a daily tech talk show, to support its mission of bringing AGI to the world and facilitate a constructive conversation about AI. TBPN will maintain editorial independence and report to Chris Lehane within OpenAI's Strategy org.

Inside Nepal's Fake Rescue Racket

198 lode 2026-04-02 12:32:23 UTC kathmandupost.com Llama Share 80

Nepal's fake rescue scam involves staging medical emergencies to claim insurance, with guides and hospitals involved in the scheme. The scam has been ongoing since 2018, with investigators finding 171 confirmed fake rescues between 2022 and 2025, resulting in millions of dollars in insurance claims.

JSON Canvas Spec

13 tobr 2026-04-02 17:32:23 UTC jsoncanvas.org Llama Share 1

Nodes are placed in ascending order by z-index, with the last node on top and the first node below others. Node types include text, file, link, group, and canvasColor.

'Backrooms' and the Rise of the Institutional Gothic

106 anarbadalov 2026-04-02 13:32:23 UTC thereader.mitpress.mit.edu Full Share 50

You don't have permission to access "http://thereader.mitpress.mit.edu/backrooms-and-the-rise-of-the-institutional-gothic/" on this server.

Significant Raise of Reports

223 stratos123 2026-04-02 09:32:23 UTC lwn.net Llama Share 110

The kernel security list has seen a huge increase in bug reports, from 2-3 per week to 5-10 per day, with most being correct and leading to faster bug fixes. This shift is expected to change the approach to security fixes, making embargoes obsolete and encouraging periodic updates.

IBM Announces Strategic Collaboration with Arm

235 bonzini 2026-04-02 09:32:23 UTC newsroom.ibm.com Llama Share 148

IBM and Arm collaborate to develop dual-architecture hardware for AI and data-intensive workloads. The collaboration aims to create flexible and scalable computing platforms for the future with greater reliability, security, and scalability.

Delve allegedly forked an open-source tool and sold it as its own

178 nickvec 2026-04-02 15:32:23 UTC techcrunch.com Llama Share 83

Delve team allegedly used Sim.ai's open source tool SimStudio without permission, violating Apache software license. Delve's founders deny allegations, but Sim.ai's CEO confirms no license agreement was in place.

Artemis II will use laser beams to live-stream 4K moon footage at 260 Mbps

193 speckx 2026-04-02 15:32:23 UTC tomshardware.com Llama Share 82

NASA's Artemis II mission used a laser-based communications system to beam 4K moon footage to Earth at 260 Mbps. The system, called O2O, will also be used for fast transmission of procedures and data.

George Goble died recently – known for first dual-CPU-Unix and fast BBQ lighting

5 finaard 2026-04-02 18:21:23 UTC legacy.com Prefix Share 1

George Harry Goble, affectionately known as "GHG," of West Lafayette, Indiana, passed away peacefully on March 18, 2026, at Majestic Care of Lafayette. He was 73 years old.Born on December 11, 1952, in Centerville, Indiana, George spent his early years there before pursuing his passion for technology. He was preceded in death by his parents, George Washington Goble and Lucille ...

Renewables reached nearly 50% of global electricity capacity last year

124 Growtika 2026-04-02 15:32:23 UTC theregister.com Llama Share 58

Renewable energy expanded by 15.5% in 2025, with solar accounting for nearly three-quarters of new capacity additions. However, the world is still behind pace to meet its renewable energy commitments by 2030.

Cursor 3

23 adamfeldman 2026-04-02 18:14:23 UTC cursor.com Llama Share 14

Cursor 3 is a unified workspace for building software with agents, bringing clarity to agent work and a higher level of abstraction. It's faster, cleaner, and more powerful, with a multi-repo layout and seamless handoff between local and cloud agents.

Sweden goes back to basics, swapping screens for books in the classroom

589 novaRom 2026-04-02 11:32:23 UTC undark.org Llama Share 304

The Swedish government is reintroducing physical textbooks and reducing digital devices in schools to improve reading and writing skills. They aim for every student to have a physical textbook for each subject, citing concerns about screen time and distraction.

Hugo's New CSS Powers

4 speckx 2026-04-02 18:17:23 UTC brycewray.com Prefix Share 0

As I mentioned in my previous post, I was intrigued when the release of Hugo v.0.158.0 introduced its css.Build function. The new powers that resulted are worth a look when you consider all the aspects of styling a site you’ve built, or plan to build, on Hugo. Still, the enhancements have certain limitations of which you’ll also want to be aware. When forming the styling structure for a ...

An Example of Statistical Investigation of the Text Eugene Onegin – Markov, 1913 [pdf]

21 jxmorris12 2026-04-02 15:32:23 UTC alpha60.de Llama Share 1

The study investigates a text excerpt from Pushkin's novel Eugene Onegin to determine the probability of a letter being a vowel. The study finds that the probability of a vowel following a vowel is higher than the probability of a vowel following a consonant.

A forecast of the fair market value of SpaceX's businesses

77 ddp26 2026-04-02 17:32:23 UTC futuresearch.ai Llama Share 98

SpaceX is targeting a $1.75 trillion IPO valuation in June 2026, but a sum-of-the-parts analysis suggests it may be overvalued by $500 billion. The valuation gap could be due to investors being overly bullish on all business segments simultaneously.

Modern SQLite: Features You Didn't Know It Had

70 thunderbong 2026-04-02 17:32:23 UTC slicker.me Llama Share 10

SQLite offers flexible data storage and querying with JSON support, full-text search, and analytical queries. It also provides strict typing, generated columns, and write-ahead logging for improved performance and reliability.

Bringing Clojure programming to Enterprise (2021)

163 smartmic 2026-04-02 08:32:23 UTC blogit.michelin.io Llama Share 95

The user designed a reference data system for manufacturing using Clojure, a dynamic functional programming language that runs on the Java Virtual Machine, due to its benefits in data manipulation and prototyping. The user found Clojure's REPL environment, code-as-data semantic, and macro system particularly useful for rapid development and prototyping, and recommends considering it for ...

Gone (Almost) Phishin'

143 luu 2026-04-02 08:32:23 UTC ma.tt Llama Share 68

A scammer targeted the user's Apple account by sending fake password reset prompts and contacting Apple Support, pretending to be the user. The scammer's convincing tactics included a fake Apple Support call and a pixel-perfect replica website to steal the user's password.

Quadratic Micropass Type Inference

13 simvux 2026-04-02 16:32:23 UTC articles.luminalang.com Llama Share 4

A new type inference algorithm prioritizes user assumptions over compiler inferences to reduce confusing error messages. It uses ordered passes to infer types in the order of user importance, with more aggressive passes at the bottom of the list.

Email obfuscation: What works in 2026?

298 jaden 2026-04-02 04:32:23 UTC spencermortensen.com Llama Share 87

Techniques to protect email addresses from spammers include HTML entities, CSS, JavaScript, and encryption, with varying levels of effectiveness. Most harvesters are unsophisticated, making even simple techniques effective, but no method is foolproof.

Telli (YC F24) is hiring engineers, designers, and more (on-site, Berlin)

0 2026-04-02 07:32:23 UTC hi.telli.com Llama Share

Businesses will increasingly use AI voice agents to interact with customers, improving the experience. AI advancements will make voice agents a no-brainer for B2C companies in the coming years.

Reinventing the pull request

66 bkolobara 2026-04-02 10:32:23 UTC lubeno.dev Llama Share 54

The author of Lubeno improved code review to help teams reduce comprehension debt by breaking down pull requests into smaller, digestible bits and allowing reviewers to focus on important changes. Lubeno's features, such as prioritizing diffs and overlaying interdiffs, aim to make code review more efficient and effective in the face of growing codebases and AI agents.

Emacs-libgterm: Terminal emulator for Emacs using libghostty-vt

68 signa11 2026-04-02 12:32:23 UTC github.com/rwc9u Llama Share 16

A terminal emulator for Emacs built on libghostty-vt offers ANSI colors, full key handling, and scrollback, but has character width mismatches and no mouse support. It requires Zig and Git installed, and can be customized with settings like shell and TERM variable.

Even GPT-5.2 Can't Count to Five: Zero-Error Horizons in Trustworthy LLMs

40 daigoba66 2026-04-02 16:32:23 UTC arxiv.org Llama Share 43

Researchers propose Zero-Error Horizon (ZEH) for trustworthy LLMs, which measures maximum problem-solving range without errors. Evaluating ZEH of state-of-the-art LLMs reveals surprising mistakes on simple problems, providing insights for safety-critical domains.

Mercor says it was hit by cyberattack tied to compromise LiteLLM

129 jackson-mcd 2026-04-02 05:32:23 UTC techcrunch.com Llama Share 37

Mercor AI startup was affected by a cyberattack linked to TeamPCP and Lapsus$ hacking groups. Mercor is investigating the incident with third-party experts and communicating with customers and contractors.

March heat in American west has left snowpack at record-low levels

74 ijidak 2026-04-02 16:32:23 UTC theguardian.com Llama Share 37

Snowpack in the American west is at record-low levels due to a historically warm winter and searing March temperatures. Experts warn that even with potential for more snow, it will be too little too late to mitigate the effects of the dry season.