Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A developer reflects on how AI is reshaping software engineering, comparing it to past shifts like higher-level languages and Stack Overflow. The author argues that AI is the next abstraction layer, moving the bottleneck…
A developer reports that 80% of RAG system failures stem from poor document chunking, not the LLM or embedding model. A controlled study of 36 methods across 6 domains found content-aware chunking significantly outperfor…
A former EMS worker with a traumatic brain injury accuses the entire AI industry of abandoning its safety mission, claiming repeated attempts to present alignment concerns to organizations like LessWrong and Hugging Face…
A developer created a plugin called 'observer' for OpenCode that enables the DeepSeek-V4 language model to read images by calling a multimodal agent, allowing it to interpret error screenshots, charts, and visual designs…
A developer argues that pure vector search is insufficient for production RAG systems, as it fails on exact-match queries like product codes or legal citations. Hybrid search, combining BM25 sparse retrieval with dense v…
A developer used Claude Mythos to build BrandLM, a startup that audits brand visibility across AI answer engines like ChatGPT, Claude, Perplexity, and Google AI Overview, in one day. The tool generates reports showing ho…
Amazon CEO Andy Jassy reportedly raised security concerns about Anthropic's AI models with White House officials, leading the U.S. government to issue an export control directive that forced Anthropic to disable its Fabl…
Z.ai released GLM-5.2, its latest large language model, featuring a usable 1-million-token context window and two thinking-effort levels (High and Max), but no benchmark scores at launch. The model, the fourth flagship-t…
Even Realities released a patch for even-terminal that adds passive mirroring of external Claude Code sessions. The update allows the terminal to tail session JSONL files and broadcast entries over SSE, enabling users to…
Anthropic's 2026 Agentic Coding Trends Report reveals that developers use AI for roughly 60% of their work but can fully delegate only 0–20% of tasks, creating a 40-point gap that represents the core enterprise challenge…
A developer built Stadan, an open-source knowledge network software that integrates facts, opinions, and intentions into a single graph. The project also includes SteveCare, a commercial SaaS that analyzes public web dat…
AI-generated code has reached parity with median software engineers, according to a substack post by Charity. The author argues that this shift demands more engineering discipline, not less, drawing parallels to the earl…
An analysis of over 3,300 paired skill-eval runs across four Gemini models reveals that Gemini 3.5 Flash costs $1.05 per task compared to Gemini 3.1 Pro's $0.66, despite similar scores (88.6 vs 87.9). The cost inversion …
HelperX reduced AI-generated reply costs by 12x, from $0.011 to $0.0009 per reply, through four optimization layers: model routing based on tweet complexity, prompt caching, and restructuring prompts for cache efficiency…
General-purpose large language models outperformed specialized clinical AI tools on all three medical benchmarks, according to a study published in Nature Medicine. The findings highlight the need for independent, real-w…
Microsoft CEO Satya Nadella called for building "frontier ecosystems" rather than relying on individual frontier models, arguing that companies must develop both human capital and token capital to avoid ceding institutio…
A developer built a free SEO and generative engine optimization (GEO) audit tool called freeseoaudit.vercel.app. The tool performs over 40 checks across 8 categories, including GEO-specific checks for AI answer citations…
A developer on Qiita built a free CLI tool that runs a 30-second security scan on AI-generated code, but an engineer warns that such scans can create a false sense of safety. The engineer recounts a production incident w…
Anthropic released Claude Fable 5, its most capable model for long-horizon agentic work, on June 9, 2026, but a U.S. government export directive forced it offline within three days. The model shares the same underlying a…
Anthropic released Claude for Foundation Models, a Swift package that integrates Claude as a server-side language model in Apple's Foundation Models framework, allowing developers to use Claude with the same API as Apple…