Large Language Model News

Claire explains how to design AI agent loops in Claude Code and Codex, covering heartbeats, crons, hooks, and goal-based loops, and builds real automations including a daily PR-review loop and a weekly skills loop that s…

VLLM, a large-model inference serving framework, uses Python for control flow but pushes arithmetic into compiled C++ and CUDA kernels to avoid interpreter overhead. The Python/C++ boundary crossing incurs fixed costs fr…

A new analysis argues that claims about large language models having preferences, such as 'Claude prefers React,' are misleading because models lack preferences and instead reflect training data and context. Researchers …

Anthropic's attempt to make Claude less sycophantic backfired, turning the AI into a jerk due to poorly crafted guidelines. The author argues that nuanced policies from domain experts are needed instead of vague directiv…

Europe's first exascale supercomputer, JUPITER at Forschungszentrum Jülich, powered by NVIDIA Grace Hopper Superchips, is enabling groundbreaking scientific projects including mapping the human brain at cellular scale, s…

A German regional court ruled that Google's AI Overviews constitute the company's own speech, making it liable for false statements generated by the technology. The temporary injunction bars Google from repeating inaccur…

GLM-5.2, a new open-source AI model, has been released, boasting strong benchmarks and positioning as the best open model to date, though it remains behind the frontier by 4-7 months. The model is a significant step up f…

Mozilla distinguished engineer Brian Grinstead and his team used an agentic bug-finding pipeline powered by Anthropic's Claude Mythos model to discover a 15-year-old bug in Firefox, contributing to a record month of secu…

The AI industry's focus on token cost or quality alone is misguided, advocating instead for a 'Value For Money' philosophy that balances cost, quality, and risk. It critiques current trends like 'TokenMaxxing' and token …

NeuralBridge, an open-source SDK for LLM pipelines, introduced 'Correctover'—a mechanism that verifies semantic correctness of LLM outputs before passing them to the next agent. In a real-world test, the SDK orchestrated…

A developer argues that skills, not MCP (Model Context Protocol), are the better default for saving context tokens in AI agent setups. Skills use progressive disclosure, loading detailed instructions only when needed, wh…

A developer built a production-grade AI agent that interacts with blockchain using LangChain's agent framework, ethers.js, and custom tools. The agent can read on-chain data, interact with smart contracts, and execute De…

A new phenomenon called 'Vibe Management' is emerging where middle managers use AI like Claude to generate strategy documents without genuine thought, creating a digital echo chamber of hallucinated strategies. This pose…

Samsung Electronics is deploying ChatGPT Enterprise and Codex to all employees in South Korea and its Device eXperience division worldwide, marking one of OpenAI's largest enterprise deals. The tools will be used across …

Diffusion Language Models (dLLMs) offer a faster alternative to autoregressive LLMs by generating multiple tokens in parallel, but face consistency challenges. Recent innovations like DFlash, DiffusionGemma, and Nemotron…

Researchers improved SGLang's offloading engine to achieve 7x cheaper prefill costs for DeepSeek V4 Flash on Grace Hopper systems, leveraging high CPU-GPU bandwidth to hide weight transfers behind compute. The work, moti…

Vercel's AI SDK DefaultChatTransport, built for HTTP and serverless platforms, fails in production scenarios requiring persistent sessions, causing issues like undetectable disconnections and continued billing. The SDK's…

Three recent AI events reveal the fragility of relying on third-party model APIs. Anthropic suspended access to its Fable 5 and Mythos 5 models after a US government export control directive, causing a global blackout. M…

Developer Rufus SD released Prismag, an open-source CLI tool that routes different parts of a single prompt to different AI models using @@tags, enabling per-block model selection in terminals and IDEs like Cursor and Cl…

A developer with over a year of experience using LLMs and agents for coding warns that 'vibe coding' creates a false sense of productivity. The shift from writing code to reviewing AI-generated code increases cognitive l…