A developer known as QLNI has built SHADOW-250M, a 250M parameter large language model that compresses to 60 MB, using a tiered context system that stores tokens older than 2048 in 1-bit form on disk,…
A latent retry bug amplified GitHub Copilot traffic from roughly 7-9K requests per second to 70-100K RPS during an outage, and Microsoft reported that a single failed request during its May 29 Azure O…
Creative technologist @stagas argues that the sunk cost fallacy no longer applies in the age of AI, as AI enables broad changes to be made quickly and effortlessly, allowing easy reversion without inc…
Marin 535B-A23B, a 535-billion-parameter mixture-of-experts large language model with 23 billion activated parameters, began open training this week on 18.75 trillion tokens across 11 NVIDIA GB200 NVL…
A developer found that connecting Claude Code to five MCP servers consumed 91,000 tokens of JSON schemas before any question was asked, with a single conversation totaling around 120,000 tokens. To ad…
A developer has launched CrowdGPT, an open-source project that enables decentralized training of large language models. The system uses a centralized server to merge client training updates into a sha…
John Abah released Abah_chat, a local-first conversational AI tool with persistent memory that stores chat histories across sessions and allows users to switch between local models without losing cont…
A CVSS 9.9 vulnerability in Cursor, CVE-2026-26268, allows attackers to execute arbitrary code on a developer's machine by cloning a malicious repository, as disclosed by Novee Security. The flaw, whi…
LifeisFair has released 'The Law of Fairness Formal Model,' a PDF document presenting a formal model of fairness. The model aims to provide a mathematical framework for defining and measuring fairness…
A GitHub user released NoBuzz (also called Claudette), a Claude Code skill that removes Anthropic's Claude's verbose, listicle-style formatting and translates responses to concise English. The tool in…
The Neural Image Format (NIF), a deep-learning-based image codec developed by Gabriel Philippus, achieves up to 26.4% lower LPIPS distortion than JPEG at 0.80 bpp on the Kodak24 dataset, with a decode…
Electrobun, an open-source desktop app framework created by Yoav Nygren, has grown to nearly 13,000 GitHub stars since its 1.x release in February, and the upcoming 2.0 version adds agentic-era featur…
Simon Willison released LLM 0.33 on August 22, 2026, as a non-prerelease version, moving the project's HTTP dependency from httpx to httpx2. The update follows a temporary compatibility fix in LLM 0.3…
Jesse Vincent reported that his agentic harness Evener, powered by GLM 5.2, autonomously developed a C compiler for ARM64 macOS in Swift that successfully compiled SQLite and passed a basic smoke test…
A developer enrolled in the LuxDev Data Science, Analytics and AI program documented their first experience using Git and SSH to publish a local project to GitHub. The project, named Kenya_Health_Reco…
Microsoft's Visual Studio 2026 v18.9, released August 11, introduces a three-level thinking effort dial for GitHub Copilot to control token usage, a Git review agent for pre-PR code review, and org-le…
A developer discovered that Ollama 0.32.14 silently falls back to CPU inference on NVIDIA GPUs with compute capability 8.6 (such as RTX 30-series and A6000) because the bundled CUDA kernels omit sm_86…
GitX, a portable Git workflow skill for AI coding agents, turns working-tree changes into logical Conventional Commits and handles branches, project checks, safe pull and push workflows, GitHub pull r…
HotCRP.com, the conference management system, now supports AI agents and bot accounts, allowing bot accounts to read submissions, set tags, and write reviews when enabled by administrators. The featur…
PromptCube, an AI developer community, argues that the most reliable way to evaluate AI communities is by the quality of reproducible code and debugging support, not follower counts or flashy homepage…