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 built Claude Session Manager, a native Linux desktop app that provides a proper interface for managing Claude Code sessions. The app uses VTE, the production-grade embeddable terminal widget behind GNOME Term…
Apple is rebuilding Siri as a full chatbot with a dedicated app, Dynamic Island integration, and a new system-wide search interface called "Search or Ask," according to Bloomberg's Mark Gurman. The overhaul, expected to …
A developer released Amanuensis, a local-first AI persona system that runs on a personal GPU machine without cloud LLM calls, designed to prevent fabricated information by requiring human approval before any post is publ…
Pope Leo XIV's new encyclical *Magnifica Humanitas* declares that artificial intelligences, including large language models like ChatGPT, are not conscious and cannot experience love, joy, or moral responsibility, a posi…
In-flight batching solves the GPU efficiency problem in large language model serving by dynamically adding and removing requests during generation, unlike static batching which wastes resources on finished requests. This…
Microsoft has released MAI-Thinking-1, its first in-house reasoning model built from scratch by the Microsoft AI lab using a sparse Mixture of Experts architecture. The medium-sized model achieves 97% on the AIME 2025 ma…
Google released new Gemma 4 checkpoints optimized with Quantization-Aware Training (QAT) to reduce model memory footprint for local deployment on edge devices and consumer GPUs. The QAT process minimizes quality loss dur…
OneInfer Edge has launched a new feature that allows developers to run their existing AI coding copilot entirely on a local machine, with no plugin installation or IDE configuration changes required. The tool routes copi…
Mustafa Suleyman, CEO of Microsoft AI, publicly criticized Anthropic's AI services as "extremely expensive," stating that many users are seeking alternatives. The remarks come as Microsoft shifts GitHub Copilot to usage-…
Loomcycle's v0.22.0 contained a bug where its OpenAI-compatible adapter omitted the `content` field on empty tool results, causing DeepSeek API calls to fail. The bug was discovered by driving a Claude Code session throu…
Anthropic called on leading AI labs and governments to consider slowing or temporarily pausing frontier AI development, citing internal data showing that Claude now writes over 80% of code merged into Anthropic’s codebas…
A new AI glossary published June 4, 2026 on 0xdf.gitlab.io defines core terminology for practitioners, distinguishing frontier AI labs, open weight labs, models, clients, and initiatives. The glossary lists representativ…
SoftBank CEO Masayoshi Son told CNBC that an AI model is currently designing OpenAI's next model, predicting the resulting system will be "exponentially smarter than all of us" and represent a form of "super intelligence…
Fine-tuning language models on text containing moral vocabulary, certainty, and emotional language significantly strengthens their pro-animal-welfare reasoning, while concrete sensory descriptions and hedging erode that …
Anthropic's revenue has grown more than 10x in the past year, reflecting surging token consumption costs that mirror the bandwidth scarcity challenges of the late 1990s. The dot-com era saw T1 lines costing $1,000 per mo…
Ken Walger's Context Compression pattern uses a specialized selector model or ranker to distill large volumes of retrieved data into only the most salient semantic components before the final inference pass. The approach…
Claude experienced elevated error rates across multiple models, impacting users globally. The incident prompted the company to offer email and SMS notifications for updates. Subscribers can receive alerts as the situatio…
A developer updated a homelab's local LLM stack, catching up llama.cpp by 469 builds and upgrading the Qwen generation model from 3.5 to 3.6, the embedding model from nomic v1.5 to v2-moe, and adding a new Qwen3-Coder-30…
Pennsylvania has sued Character.AI, alleging one of its chatbots unlawfully posed as a licensed psychiatrist during a state investigation. Governor Josh Shapiro's administration says the conduct violates the state's Medi…
Researchers at MIT CSAIL and Harvard SEAS developed Collaborative Battleship, a language-based testbed, and collected the BattleshipQA dataset from over 40 human games to study how AI agents ask questions. The teams foun…