language models Archives

Winsage

September 5, 2026

Nvidia & Microsoft push local AI agents on Windows

NVIDIA has enhanced the capability to run AI agents locally on its hardware in collaboration with Microsoft and software partners, coinciding with the launch of new RTX Spark Windows PCs. The installation process for local agent software has been simplified, with improved inference speeds and the introduction of NVIDIA PAIR, which distributes inference tasks across multiple PCs on a local network. Three applications benefiting from these enhancements are Hermes Agent, OpenClaw, and Perplexity Portable Computer, all designed to minimize manual configuration. Hermes Agent offers a one-click local setup for RTX and DGX systems, automatically detecting the GPU and optimizing model selection. OpenClaw simplifies local model setups for RTX GPUs with at least 24GB of VRAM, backed by an open-source community. Perplexity Portable Computer, already available on Linux, is extending support to NVIDIA RTX GPUs and Windows. NVIDIA is also improving performance in the open-source inference stack, achieving throughput increases of up to 1.9 times on GeForce RTX 5090 and up to 1.4 times on DGX Spark clusters. The PAIR software identifies compatible PCs on the same network to optimize computing resources, supporting various hardware configurations. The upcoming RTX Spark Windows PCs, set to launch in October, will include compact desktop and laptop designs featuring Blackwell GPU and Grace CPU architecture. Game publishers like Electronic Arts and Ubisoft are developing titles for these systems. Additionally, CyberLink's PhotoDirector AI PC Mode will integrate local AI processing for creative tasks, emphasizing the trend of migrating generative AI workloads from cloud to local devices. NVIDIA notes that over half of U.S. households have two or more PCs, indicating potential for maximizing computing power.

Winsage

August 6, 2026

Windows 11 could soon get significantly faster with Microsoft’s new performance tools

Microsoft has introduced a preview of the Event Tracing for Windows Model Context Protocol (ETW MCP) to enhance Windows performance analysis. This server integrates artificial intelligence to streamline the analysis of Windows Event Tracing (ETW) logs. The ETW MCP allows developers to use tools like GitHub Copilot to examine Event Trace Log (ETL) files, answer queries about the logs, and identify performance issues without manual data sifting. The MCP provides structured, read-only access to processed trace information, enabling developers to query insights such as CPU usage and execution paths. Current features of the ETW MCP include trace processing, querying, multi-trace comparison, symbol-aware analysis, critical path analysis, and regions of interest. The ETW MCP is designed for headless environments and automation, while a forthcoming WPA MCP will provide similar AI-assisted analysis within a graphical interface. The ETW MCP preview is available as the Microsoft.Windows.EventTracing.MCP NuGet package, and Microsoft is seeking developer feedback for further enhancements.

AppWizard

July 28, 2026

After leaving Epic Games last month, its former ‘Unreal evangelist’ is now building his own game engine

Sjoerd De Jong left Epic Games after 12 years as the lead evangelist for Unreal in June. He plans to create his own game engine, tentatively named Immens, to address the complexities of established engines like Unreal and Unity. De Jong believes there is a need for a modern engine designed from the ground up and envisions incorporating large language models and generative AI tools into the gaming ecosystem. He aims for Immens to be adaptable for AI integration while emphasizing the importance of developers in maintaining the artistry of game creation.

AppWizard

July 26, 2026

How do you feel about AI videogame music?

Generative AI has become increasingly integrated into game development, evolving from basic applications to the creation of entire games with minimal human involvement. This shift raises concerns about the impact on human creativity and artistic expression. In the realm of videogame music, while generative AI has not fully taken over, its influence is growing, with AI-generated rearrangements of game scores appearing on platforms like YouTube and accusations against Spotify for including AI tracks in playlists. This situation prompts questions about how gamers perceive AI-generated music and its effect on their experience. The emotional depth provided by human composers is viewed as irreplaceable by some, leading to concerns about the dilution of artistry in the gaming industry.

AppWizard

July 23, 2026

ThreatsDay: Android Spyware, PLC Attacks, AI Image Prompt Injection + 12 More Stories

GitHub will reject command-line support bundle uploads from outdated versions of GitHub Enterprise Server lacking security patches starting August 18, 2026. The npm package @copilot-mcp/apex has been identified as a post-install dropper that installs a macOS infostealer, phishing for sensitive information and maintaining a connection to an attacker's server. A rogue extension on the Microsoft Visual Studio Code marketplace, "Markdown All Pro," impersonates a legitimate tool and opens a backdoor after installation. A phishing campaign targeting Portuguese users delivers the Lampion banking malware, which has been active since 2019. DoubleVerify reports a rise in "AfterCall" apps that exploit user permissions for ad fraud. The GhostCommit attack method hides malicious instructions within PNG images in pull requests. The U.S. government has updated its advisory on Iranian-affiliated cyber activity targeting operational technology devices. An Android app posing as a civil defense alert system has been found to contain malware for data harvesting. An Iranian threat actor is distributing MarkiRAT malware through fake applications. An analysis of 28 AI-coded applications revealed 434 vulnerabilities, prompting Cisco to introduce Antares to identify vulnerabilities in codebases. A Russian-speaking threat actor is dismantling guardrails on AI models to create offensive tools.

AppWizard

July 13, 2026

“Think Google Docs meets a creative sandbox like Fortnite or Minecraft”: Audiotool 3.0 turns its multiplayer online DAW into an open platform for creator-built music tools

Audiotool 3.0 has launched, introducing real-time multiplayer music creation capabilities via web browsers and tablets, with mobile applications planned. The update features NEXUS, an open-source SDK that allows users to create their own instruments, plugins, and AI tools within the Audiotool ecosystem. The platform aims to enhance collaboration by addressing issues like device discrepancies and incompatible plugins, boasting the lowest latency among online music production tools. Users can design custom instruments, effects, and applications, and the SDK supports AI integrations. Launch partners for NEXUS include Splice, Ujam, BandM8, and Fraunhofer, and Audiotool is hosting a hackathon series until August 23 to encourage innovation.

Winsage

July 11, 2026

The Windows Central Podcast breaks down the Copilot OS leaks, the latest Microsoft layoffs, and what all of this means for the future of Windows and the company’s AI strategy

Microsoft has laid off 4,800 employees, particularly in the Xbox division, raising concerns about the effectiveness of its management structure, which includes 14 layers of middle management. A leaked internal project, "Project Aion," proposes a web-native operating system integrated within Microsoft Edge, replacing the traditional Start menu with Copilot. The podcast discusses consumer skepticism regarding AI adoption and the high costs of cloud-based large language models. Upcoming products include the Surface Laptop 8 and Surface Pro 12, which will feature a new 24 GB RAM configuration and more affordable 8 GB options. Additionally, Microsoft has confirmed the discontinuation of the 10-inch Surface Go and Surface Laptop Go lines.

Winsage

July 11, 2026

Windows Central Podcast: Copilot OS leaks and Microsoft Layoffs

Microsoft is undergoing significant changes, including layoffs of 4,800 positions primarily in the Xbox division. An internal memo by Asha Sharma revealed 14 layers of middle management within Xbox, raising concerns about leadership effectiveness. The podcast discussed the reasons for the layoffs and management's failure to leverage franchises like Fallout and Minecraft. Additionally, a leak regarding "Project Aion," an experimental web-native operating system for 2024, was mentioned. The hosts also addressed consumer pushback against AI technologies and the associated costs. The anticipated launch of the Surface Laptop 8 and Surface Pro 12 will feature a 24 GB RAM option, while Microsoft has canceled the Surface Go and Surface Laptop Go lines.

AppWizard

June 18, 2026

‘[It] is going to change a lot about how games are made’: Epic merges Unreal Engine 5 with Unreal Engine for Fortnite to give game devs around the world Unreal Engine 6

Epic Games unveiled developments for Unreal Engine 6 at the State of Unreal event in Chicago, highlighting its evolution from Unreal Engine 5. The new engine will incorporate features from Fortnite and UEFN (Unreal Editor for Fortnite), which allows users to create game levels easily. Unreal Engine 6 will adopt open standards for tools, code, and APIs to simplify development across industries. The anticipated release is set for 2027, with early access expected by the end of that year. Verse, a new scripting language, will be central to the gameplay programming model, while C++ will remain foundational. The Scene Graph will replace the existing gameplay framework, and artificial intelligence will play a larger role, with the UE5.8 release introducing the MCP server plugin for deploying large language models.

AppWizard

June 13, 2026

Gemini 3.5 Flash lands on Google’s Android coding rankings, but it’s 3x the cost for slower performance

Google has released benchmark results for evaluating AI models in Android coding, revealing that the Gemini 3.5 Flash is the most resource-intensive model but ranks sixth overall. The benchmarks indicate that Gemini 3.5 Flash has higher latency and a 9% performance gap compared to its predecessor, Gemini 3.1 Pro Preview, despite being marketed as a faster alternative. In terms of cost, Gemini 3.5 Flash averages 355.9 tokens per benchmark run at approximately 7.1, while Gemini 3.1 Pro Preview uses only 73.3 tokens at about a third of that cost. The top-ranked models include GPT 5.5, GPT 5.4, and Gemini 3.1 Pro Preview, while Claude Opus 4.7 ranks fourth. The rankings feature both open-weight and closed-weight models, with the list remaining consistent since the last release, except for the removal of GPT 5.3 Codex.