Full-text search across 63921 articles. Combine with topic and date filters; results sorted by relevance.
RunInfra enables running Kimi-Linear-48B, a distilled version of the full 2.78-trillion-parameter Kimi K3 model, on a single consumer GPU such as the RTX 5090 with 32 GB VRAM, achieving 113.83 tokens per second at 3.18x …
A developer has released claude-code-merge-queue, a TypeScript tool that serializes commits from parallel Claude Code agents on a single machine, after running four to five agents pushing up to 90 commits a day through 8…
Developer ndom91 released open-plan-annotator, a fully local annotation UI plugin for coding agents including Claude Code, OpenCode, and Pi, that intercepts plan mode and lets users mark up plans with strikethrough, repl…
Bayu, founder of NAEOS, argues that the next evolution in AI-assisted software development is not prompt engineering but AI engineering. He introduces NAEOS (Nusantara AI Engineering Operating System), a foundation layer…
Developers are rejecting proprietary wrappers around coding agents and demanding raw access to the agents they already use, according to an analysis of buying conversations. The article argues that hard-coded determinism…
Palo Alto Networks acquired Koi, an AI-native developer security product, to address blind spots in EDR/XDR systems that cannot detect threats from VS Code extensions, local MCP servers, and rogue AI coding assistants. I…
A guide for Python beginners identifies seven common mistakes that slow learning, including jumping between tutorials, passive watching without coding, mixing tabs and spaces, confusing 'is' with '==', using mutable defa…
Thoughtbot reports that startup founders using AI tools like Claude, Lovable, and Bubble.io can prototype quickly but still face challenges in product decisions, code validation, and strategic planning. Non-technical fou…
A user who tested nearly every major AI model subscription reports that ChatGPT's $20 plan offers the best value, providing generous access to GPT 5.6 Sol, a state-of-the-art model competitive with the best in the indust…
Theori's Xint team reported 434 code-validated security findings across 28 applications built or rewritten with Anthropic and OpenAI coding models, with resource-exhaustion and denial-of-service flaws leading at 93 findi…
Google's Gemini 3.5 Pro flagship model has missed its public launch deadline three times, with a late-June training data update targeting coding failures producing disappointing results that made things worse, according …
Korean AI startup VIDRAFT has released Aether-7B-5Attn, a 6.59B-parameter Mixture-of-Experts foundation model on Hugging Face under Apache-2.0. The model features five heterogeneous attention mechanisms arranged via a 7×…
In November 2025, Anthropic released Claude Opus 4.5, a model that made AI agents capable of autonomous task execution, leading to a dramatic shift in software development and personal productivity. Nat Friedman of Meta …
A developer found that their 14,000-line Python desktop app, built with heavy AI assistance, suffered from 'diff debt'—code merged without being read. The clearest example was five separate PDF generation modules failing…
Block launched Buzz on July 21 as a free, open-source workspace where people and AI agents share chat channels, code repositories, and automated workflows. The Nostr-based system gives agents separate cryptographic ident…
A hobbyist game developer reports that after two weeks using Claude Code on two small game projects, the AI tool adds excessive text to screens, wastes tokens on screenshot verification, and struggles to automate the sub…
A senior engineer quit after spending eight months reviewing AI-generated code instead of building, telling her manager she was burned out from being a 'verification layer for an agent.' The manager, who had celebrated v…
A new benchmark and tutorial series from developer dotnetspark reveals that model-routing systems optimize the wrong axis by routing queries to models without grounding, finding that a cheap local model (phi3:mini, ~3.8B…
Local LLM usage data is already stored on users' disks via client transcripts, but plan limits remain server-side and unreadable, forcing any tool that converts consumption into a percentage to estimate rather than measu…
Meta executives have spent weeks arguing that its NameTag face-recognition feature for Ray-Ban smart glasses does not exist, despite WIRED finding code for the feature in the Meta AI app as early as January and Meta CTO …