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 at Riversea Lab accidentally ran Claude Code with --dangerously-skip-permissions on a production database after misconfiguring an environment file. The multi-agent loop hit 12K KV writes per minute due to a c…
A developer built NoFlattery, a single-file React SPA that hosts a council of AI agents with different reasoning biases to debate user decisions. The app generates a Decision Record with a verdict, reasoning, risks, and …
Subrata Kumar Das published a lesson on Generative AI as part of a new course, explaining the evolution from traditional rule-based software to modern AI agents. The lesson covers the shift from predictive AI to generati…
A developer tracked writing output over 90 days and found that switching from Obsidian-first drafting to Claude-first drafting nearly doubled usable words per 90-minute block, from 420 to 780. However, the AI drafting in…
OpenCode, an open-source AI coding agent, surpassed 160,000 GitHub stars and now has 172,000+ stars with 7.5 million monthly active developers. It is model-agnostic, supporting 75+ providers, and features LSP integration…
Multiple independent researchers are seeking endorsements on arXiv to submit papers on machine learning topics, including a discrete choice model for LLM alignment, a control-theoretic approach for long-form generation, …
Several authors, including Anthony Horowitz, are openly using AI tools like ChatGPT for research and writing assistance, challenging the stigma against AI in publishing. A Cambridge study found that a third of UK novelis…
Chinese AI developers are accelerating development of trillion-parameter foundation models to narrow the gap with US rivals, as Washington imposes export controls on leading US AI software. DeepSeek launched its first tr…
Zhipu AI's GLM-5.2 open-weight model has passed multiple real-world 'vibe checks,' with Jeremy Howard and Artificial Analysis rating it comparable to GPT 5.5 and Opus 4.8, marking a shift from benchmark-only open models.…
Generative AI is transforming .NET development by automating boilerplate code, bug prevention, documentation, unit testing, and log analysis. Tools like GitHub Copilot, Tabnine, Microsoft IntelliCode, OpenAI Codex, and D…
A senior software engineer describes how LLM-assisted coding has rekindled their engagement and productivity, leading to a surge in shipping personal and professional projects. While they find the experience reinvigorati…
A developer outlines the repetitive infrastructure—auth, billing, AI features, and deployment—required for every AI SaaS product, arguing that the real challenge is turning an AI API call into a monetizable product. The …
A developer built a personal operating system for Claude Code, an AI coding assistant, to address recurring issues like misunderstood requirements, out-of-scope modifications, and false completion claims. By adding struc…
Z.ai (Zhipu AI) co-founder Tang Jie declared on June 18 that China will reach Fable-class AI capability before 2027, countering Elon Musk's estimate of Q1 2027. The prediction follows a U.S. Commerce Department order on …
Boston Children's Hospital and OpenAI used AI to diagnose 18 children with rare diseases that had stumped doctors, identifying genetic causes in 5% of previously unsolved cases. The findings, published in NEJM AI, demons…
Researchers introduced SEVRA, a serving-layer controller that decides when a frozen reasoning model should verify its answer instead of thinking longer, finding that selective verification improves accuracy while reducin…
A mathematician predicts that AI tools like Claude and ChatGPT will transform mathematical research by enabling rapid testing of computational guesses, completion of partial proofs, and collaborative problem-solving. The…
A developer A/B tested LLM silence by toggling an embodiment system prompt with the user prompt "Be silence," finding that the present prompt produced 0 bytes while its absence yielded speech, using GPT-5.2 and Claude Op…
Profile v2.1.4, a physics-aware optimizer for vLLM inference servers, achieved a 15x throughput increase from 31 to 470 tok/s and a 93% cost reduction on an NVIDIA A100 GPU. The tool uses roofline math to compute theoret…
A developer built a groundedness-checking layer for a local RAG system to catch hallucinations, running on Ollama with an 8B model. The layer decomposes answers into claims and verifies each against source passages, catc…