Ars Technica3 days agoSoftware Engineering
Nvidia fixes the 8GB RAM problem with one of its GPUs—if you can pay for it
Framework charges nearly double for the 12GB version of the mobile RTX 5070.
👍 Like💬 Comment↗ Share
New site is out yeyyyyyy
Stories in this category.
Framework charges nearly double for the 12GB version of the mobile RTX 5070.
Study authors say courts should reconsider rulings in light of this new evidence.
Roku’s $2.99 streaming service Howdy has topped 1M subscribers, showing demand for cheaper, low-commitment alternatives to pricier streamers.
VoidZero has released an experimental Angular compiler in Rust, promising improved build performance compared to Angular CLI. Benchmarks indicate it is up to 6.4x faster than Angular CLI, especially for larger applications. Developed over two months with AI collaboration, it integrates with Vite and offers significant reductions in build time. The initiative is for research purposes only. By Danie
Microsoft has released .NET 11 Preview 3, the third preview of its upcoming Standard Term Support version due in November 2026. The update spans the runtime, SDK, libraries, ASP.NET Core, EF Core, and .NET MAUI, with Runtime Async leaving the preview-feature gate, new JIT optimizations, Zstandard compression in ASP.NET Core, and expanded MAUI Map controls. By Almir Vuk
Tracy Bannon shares a cautionary tale of "The Sorcerer’s Apprentice" to illustrate the risks of unbridled AI autonomy. She discusses the shift from bots to autonomous agents, explaining how reckless speed leads to "Architectural Amnesia." She provides a concrete framework for "Minimum Viable Governance," focusing on identity, delegation, and ADRs to manage debt at machine speed across the SDLC. By
Manufacturers backed effort to repeal the law but ultimately failed.
The object will be traveling at 2.43 km a second, or 5,400 mph, upon impact.
Sauce Labs has announced the general availability of Sauce AI for Test Authoring, an AI-driven agent designed to translate business intent directly into executable test suites, marking a shift toward what the company calls Intent-Driven Testing. By Craig Risi
On sale through the NASA exchange.
Mistral AI has launched Workflows, an orchestration layer for enterprise AI that is now in public preview. This release addresses a significant challenge as AI models and agents become more advanced, while reliably deploying them in production remains difficult due to a lack of infrastructure for coordination, monitoring, and recovery. By Robert Krzaczyński
The schedule for QCon AI Boston 2026 (June 1-2) is now live. The two-day program groups sessions around context engineering, inference economics, agent reliability, and how AI is changing the software development lifecycle. Speakers include engineers from DoorDash, LinkedIn, Netflix, Apple, and Red Hat. By Artenisa Chatziou