Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
DeepMind released DiffusionGemma, a new large language model that uses discrete diffusion to generate entire sequences in parallel instead of autoregressive token-by-token generation. The model achieves over 1,000 tokens…
A developer testing Freebuff on an enterprise-grade Laravel POS system found the AI tool excels at code review—catching bounded context violations, race conditions, and PgBouncer compatibility issues—but produces dangero…
South Korean AI silicon startup Rebellions is prioritizing memory-centric architecture in its technology roadmap, leveraging strategic ties with SK Hynix and Samsung Foundry as it explores IPO options. The company's seco…
A simulation-based study published in the Journal of Medical Internet Research found that patient cognitive bias reduced large language model diagnostic accuracy by 10 to 40 percentage points across six models tested on …
A former xAI engineer, Devin Kim, sued the company and its parent SpaceX in California state court, alleging he was wrongfully terminated in September 2025 after repeatedly warning supervisors that the Grok chatbot posed…
Microsoft is restricting employee access to Anthropic's Claude Fable 5 after Anthropic introduced mandatory data retention rules tied to its safety classifiers, according to The Verge. Fable 5 is absent from the model pi…
Mixture-of-Experts (MoE) architecture allows large language models to use only a fraction of their total parameters for each token, enabling models with 671 billion total parameters to run at speeds comparable to a 37-bi…
A critical Remote Code Execution (RCE) vulnerability in AMD hardware remains unpatched after the vendor declined to fix it, leaving users exposed to potential full system compromise. GitHub has deployed LLM-driven verifi…
Anthropic has become the most valuable AI startup in the world following a $65 billion Series H round that valued the company at approximately $965 billion, surpassing OpenAI's valuation. The company's revenue has crosse…
Anthropic's Claude Code now integrates OpenAI's Codex CLI in a three-phase workflow where Claude plans and reviews while Codex executes tasks, burning ChatGPT usage limits instead of Claude limits. The system, available …
A developer achieved 775 tokens per second running the full BF16 DiffusionGemma model on an Nvidia RTX 6000 Pro using a Red Hat fork of vLLM, demonstrating extremely fast local AI inference at short contexts. However, pe…
A developer has released nano-agent, a lightweight shell agent written in Rust that interfaces with any OpenAI-compatible API to execute commands with user approval. The tool, available via Cargo install, offers both one…
Benchmarks comparing AI models consistently pit Fable 5 against GPT 5.5 xhigh, but a user argues this is an unfair matchup because GPT 5.5 Pro, which has been available for some time and outperforms Opus 4.8 Max, would b…
OpenAI has appointed Thibault Sottiaux as head of core products, putting him in charge of transforming ChatGPT into a personalized "super app" that handles tasks across users' personal and professional lives. Sottiaux, w…
The overuse of Tailwind CSS templates in AI-generated web applications has created a recognizable "slop" aesthetic that signals rushed, vibe-coded products to users. A search of Hacker News "Show HN" posts found the firs…
A California mother is suing OpenAI, alleging that "deliberate design decisions" in its ChatGPT-4o model contributed to her 24-year-old daughter's suicide in July 2024. The lawsuit, filed in San Francisco County Superior…
A new experiment testing Anthropic's Claude Fable 5 model on a Next.js coding task revealed that while the model successfully identified and fixed a cache invalidation bug, its failure modes have become more subtle compa…
GKE Labs has launched OpenRL, an open-source, self-hosted training API for fine-tuning large language models on Kubernetes. The API decouples post-training infrastructure from reinforcement learning research workflows, u…
Amazon published a how-to demonstrating an intelligent document processing pipeline on Amazon Bedrock that combines on-demand FIFO queue processing for low-latency single documents with batch inference for asynchronous, …
Paramount streaming leaders reported sharp productivity gains after adopting AI tools, including Anthropic's Claude for coding work, according to internal materials and employee accounts reviewed by Business Insider. An …