Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
CEOs of OpenAI, Anthropic, and xAI have paused development efforts over concerns about the rapid advance of AI capabilities, Axios reported, a slowdown that has pushed the market-implied probability of Anthropic reaching…
A developer outlined a reusable architecture for production AI applications on AWS, arguing that a multi-tenant retrieval-augmented generation system should be treated as a distributed system with an LLM inside it rather…
Microsoft's AI Foundry model catalog now serves DeepSeek v4.1 Flash, which launched this week and drew 993 points and 565 comments on Hacker News, as the fastest Azure-native way to deploy the open-weight model behind a …
The AllSpark team released Iris-mini and Iris-pro, two open-source search agents built on Qwen models that lead benchmarks among open-weight models in their size classes, according to the accompanying paper. The training…
Windsurf's Flow feature can orchestrate a pipeline that converts markdown text into an HTML/CSS presentation framework such as Reveal.js to build AI slide generators, according to a first-person account of the workflow. …
Google moved its roughly 90-person AI responsibility team out of Google DeepMind and into its global affairs division, the unit handling policy and lobbying, completing the transition on September 1, as reported by The W…
Researchers at Apeiron Intelligence, the Artificial Intelligence Research Institute (IIIA-CSIC) and Universitat Politècnica de Catalunya (UPC) published a tutorial introducing an operator-level framework that classifies …
AgentJIT, a new open-source Python library, compiles dynamic LLM agent workflows into deterministic Python code that runs in under 0.1 milliseconds with zero LLM tokens consumed on repeat executions, according to its Sho…
A developer released ThreadShelf, a local-first tool that archives and semantically searches AI chat exports from six providers: ChatGPT, Claude, Google AI Studio, OpenRouter, LM Studio, and Grok. The archive pipeline — …
A second-part technical article on latency optimization for open-weight LLM inference details four techniques measured on a fixed deployment of Qwen3-8B (Apache-2.0) served on vLLM through SageMaker's DJL-LMI container, …
China's average daily AI token calls passed 140 trillion by late March 2026, up from 100 billion at the start of 2024 and 100 trillion at the end of 2025, according to Xinhua-cited figures, and a National Data Administra…
A developer has proposed a "Markdown Blackboard" pattern for coordinating multiple autonomous AI coding agents through a single shared Markdown file in a project workspace, using checklists and inline notes as a state ma…
Hugging Face released Transformers v5.17.0 on September 9, adding seven model architectures in a single minor release, including Tencent's 780-billion-parameter mixture-of-experts model and Moonshot AI's Kimi Linear hybr…
OpenAI has delayed its planned initial public offering beyond 2026, with chief executive Sam Altman telling Fortune there is still "a lot of stuff to do" around safety, alignment and government-industry cooperation. Altm…
Chinese president Xi Jinping unveiled a Brics AI Open Source Zone and a Brics Special Economic Zone partnership at the Brics summit in Delhi on Sunday, saying China would lead cooperation on large language models, traini…
Promptfoo 0.123.0 can be wired into GitHub Actions via promptfoo/promptfoo-action@v1 to score prompts against fixed test cases on every pull request, failing the PR when an edit breaks tone, policy rules, or output shape…
A technical explainer derives the probability of a large language model trajectory from first principles using the chain rule, factorizing p(τ|θ) into the initial state distribution p(s0), policy terms πθ(at|st), and env…
A developer built Glacier.Inference, a ~15 MB single-file Native AOT inference engine written in pure C# on .NET 10 that runs GGUF models directly across NVIDIA, AMD, and Intel GPUs without C++ binaries or the CUDA runti…
A developer published a beginner's guide explaining retrieval-augmented generation (RAG) through five progressive levels of building an AI assistant that answers questions from a user's own documents rather than only its…
OpenAI launched its new cutting-edge model, GPT-6 Astra, which the company says gives users more control of their computer and removes limitations on tasks that can only be performed via a GUI, with the model slowing dow…