The tracker detects statistically significant degradations in Claude Code with Opus 4.5 performance on SWE tasks. It uses daily evaluations with 50 test instances and aggregates weekly and monthly results for reliable estimates.
Google is releasing Project Genie, an experimental web app that lets users create and explore interactive worlds using Genie 3, a world model that simulates real-world scenarios. The app allows users to create characters, worlds, and define how to explore them, with features like real-time path generation and world remixing.
Haakam, Michael, and Adi built AgentMail, an email inbox API for agents, providing a universal protocol with identity and authentication. AgentMail allows developers to create email agents that can autonomously complete tasks and interact with users through email.
Researchers tested 14 frontier LLMs on 23 OpenTelemetry tasks across 11 languages, finding most struggled with complex tasks. The cheapest and fastest model, Gemini 3 Flash, performed well but still failed on many tasks.
Websites use necessary cookies for security and functionality, and analytics cookies to measure usage. Other cookies help with advertising, targeting, and performance.
The Meteosat Third Generation-Sounder satellite has shared its first images, providing data on temperature and humidity for more accurate weather forecasting over Europe and northern Africa. The satellite's Infrared Sounder instrument captures data on temperature and humidity, enabling the creation of three-dimensional maps of the atmosphere.
Google Labs introduces new features: Genie takes over hero, Pomelli moves to featured component, and CC is an experimental AI agent in Gmail. Mixboard and Learn Your Way are also experimental tools for idea exploration and personalized learning.
0
2026-01-29 17:06:05 UTC
ycombinator.com
Llama
Reflex is a unified platform for building, deploying, and managing applications end-to-end. It replaces the fragmented enterprise stack with a single platform, empowering teams to own the entire lifecycle of their apps.
The US Cybersecurity and Infrastructure Security Agency's acting head uploaded sensitive government files into ChatGPT, triggering security alerts and a federal review. The incident raises concerns about sensitive data leaving internal networks as the US government pushes AI adoption across federal agencies.
CERN's Large Hadron Collider is reusing hot water to heat homes in Ferney-Voltaire, reducing CO2 emissions by thousands of tonnes. The heat exchange system supplies up to 5 MW of heat to the local area, with plans to double this amount in the future.
The user tested various terminal themes including macOS Terminal.app, Tango, and Solarized to find suitable colors for syntax highlighting.
The user discusses the benefits of 3D printing and software development, highlighting the ability to create niche solutions tailored to specific needs. This allows for efficient problem-solving and customization, making it a unique advantage in both 3D printing and software engineering.
Apple has set a new deadline of November 1, 2026 for Patreon creators to switch to the App Store's in-app purchase system, with Apple taking a 30% commission on digital goods. Patreon creators can sidestep the commission by completing payments via the platform's website.
Cloudflare has developed Moltworker, a middleware that allows running Moltbot, an open-source AI agent, on Cloudflare's Sandbox SDK and Developer Platform APIs. Moltworker uses Cloudflare's features like Browser Rendering and R2 for persistent storage.
EmulatorJS is a rewritten project with a new version system and CDN for cores and files. Use releases or CDN for production, and consider funding to remove demo ads.
Scala received a €377,300 investment from the Sovereign Tech Fund to strengthen its security, maintenance, and developer experience. The investment will be used over two years to improve Scala's core foundations, tooling, and documentation.
Researchers found vulnerabilities in widely used NFC technologies, allowing attackers to recover cryptographic keys through relay attacks and partial key overwrites. Proper configuration and key diversification can mitigate these issues, but many deployments are at risk due to configuration lapses and use of non-NXP compatible cards.
The AI surge is causing DRAM producers to focus on HBM, potentially killing small VPS hosting businesses like mine due to high costs. This mirrors the 2000s when small ISPs died due to Big Telecom lobbying and lack of line sharing.
Researchers implemented a Go-playing neural network on an 8-bit Motorola 6809 microprocessor, achieving strength comparable to GNU Go. The software was run on a Thomson MO5 microcomputer in 2023.
Mark Surman, Mozilla's president, is building a "rebel alliance" of tech startups and developers to make AI more open and trustworthy, countering industry heavyweights like OpenAI and Anthropic. Mozilla is investing in mission-driven tech businesses and nonprofits to promote AI transparency and counterbalance the power of large companies.
Beautiful-mermaid is a fast and beautiful text-based diagram renderer that works everywhere, from rich UIs to plain terminals, and is based on the Mermaid standard. It has a powerful theming system that allows for easy customization of diagram colors and can also use VS Code themes directly via Shiki integration.
Mental arithmetic in grade school (e.g., memorizing your times tables) is typically justified on the grounds that facility in basic calculations makes it easier to focus on higher-level problems that require being able to do these manipulations. When working on DTensor, I have also found it important to be able to quickly calculate what shardings you get when you do matrix multiplies on ...
Test and explore the Shaped query language with real data
Waymo's robotaxi struck a child at 6mph after braking from 17mph, with the child suddenly entering the road from behind a stopped SUV. The NHTSA is investigating the incident, which occurred near an elementary school during drop-off hours.
Taking 5000 IU of Vitamin D and 1500 mg of 60%-EPA Omega-3 supplements daily may be effective in improving mental health, with Vitamin D being at least twice as good as antidepressants. These supplements are cheap, safe, and over-the-counter, and can be taken in addition to existing antidepressants, but it's recommended to consult a doctor before starting.
A campus email system experienced a problem sending emails over 500 miles due to a Sendmail version downgrade caused by a server upgrade. The issue was resolved by identifying the incorrect Sendmail version and updating it to the correct one.
The Comet is a highly customizable, all-magnetic device with 40 IO pins and expandable storage up to 2TB. It can run ARM-based packages and supports various extensions for mobile data, calling, and decentralized communication.
Virginia "Ginny" Oliver, Maine's "Lobster Lady," died at 105 after a 97-year lobster fishing career. Her remarkable life inspired tributes from Governor Janet Mills and celebrities like Mark Hamill, who celebrated her tenacity and spirit.
apt-bundle is a tool for managing Debian-based system dependencies. It uses an Aptfile to define dependencies and can install them with a single command.