Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Advocate General Maciej Szpunar advised the Court of Justice of the European Union on November 3 that Google's Gemini chatbot may infringe EU copyright law when it reuses journalistic articles to generate answers, though…
A developer ran 586 traces of a LangGraph folder-research agent against AgentPulse, a metadata-only observability platform, using qwen2.5:7b on an RTX 4060 and a Pixel 8 Pro running a 3B model via llama.cpp in Termux. At…
Researchers introduced ParetoSaddler, a Pareto-guided method for optimizing LLM agent harnesses jointly on response quality and cost without modifying model weights. The work addresses a limitation in existing harness-op…
A paper titled "SpecPilot: Autonomous Algorithm Discovery and Systems Optimization for Speculative Decoding" questions whether released parallel-drafting systems achieve the full speedup that parallel drafting permits fo…
A quantized weight set for Qwen-Image-2.1-Turbo, a 7B text-to-image model, has been published on Hugging Face by the maintainer Abenzerps, enabling local image generation in ComfyUI, stable-diffusion.cpp and MFLUX. The r…
A developer built an AI-powered AP Lang essay grader that scores student essays against the College Board's official six-point rubric, finding that the model itself was the easy part and that calibrating it to grade like…
A developer outlined an "envelope" design for unattended LLM pipelines in which the model may only choose among human-approved plans, with failure feedback carried forward as typed template messages rather than a growing…
A developer published a walkthrough for building a multi-agent invoice-triage system using LangChain's Deep Agents framework and the Model Context Protocol (MCP). The architecture pairs an orchestrator that plans and mai…
Brazil's Superior Electoral Court ordered Meta on Oct. 15 to treat President Luiz Inácio Lula da Silva and opponent Flávio Bolsonaro equally under threat of a daily fine of 500,000 reais (about $98,000), after Lula's cam…
A developer benchmarked Google's Gemini 4 Argon against Claude Opus 5.5 inside the OpenCode coding agent across six real coding tasks, grading each run with hidden tests. Both models passed every hidden test in all 24 ru…
OpenAI's GPT-6 Astra, released on 3 September as a frontier model for computer use and coding, outperformed all 40 trained robot policies on RoboDojo's benchmark of 42 simulated tasks with no robot-specific fine-tuning, …
Research firm SemiAnalysis published GDPx, a benchmark of 206 real paid client tasks across 11 professional domains contributed by 14 partner organisations, and found that the best of 20 frontier models completed only 12…
A developer published a walkthrough for building a command-line retrieval-augmented generation Q&A bot over internal company documents using Pinecone's Python SDK v10.3.0 and Anthropic's anthropic 2.4.0 with the claude-s…
An endpoint tracking page for z-ai/glm-5 served via SiliconFlow at fp8 reports a B3IT (border-input) change with a total-variation distance of 0.52, while logprob tracking is unavailable because the endpoint returns no l…
A Hirundo study found that GLM 5.3 Flash, tested through OpenRouter, wrote account-recovery code for a mock repository named after Revolut that was 28% more likely to allow account takeover than code written for a Chines…
A report published on RikiyaT's blog and covered by Arcturus Labs finds that frontier language models can read, reason about, and even update neural network weights presented as plain text, identifying a Pac-Man-playing …
A developer outlined five engineering practices for making AI agent workflows production-reliable, centered on separating LLM interpretation from deterministic action execution. The approach has the model return a struct…
A developer built a free AI text humanizer in vanilla PHP 8 that chains two free LLM providers, OpenRouter and SiliconFlow, so that a failed, rate-limited, or empty response from the first engine is transparently retried…
Hystax open-sourced OptScale Kompressor, a Rust binary that strips machine-generated noise from coding agents' context before it reaches the LLM API, cutting the provider bill by 22.7% in the company's Terminal-Bench 2.1…
A November 1, 2026 arXiv paper by Suze Van Adrichem introduces "hinted scaling laws," which use privileged information to create a continuum of easier benchmark versions so small models score above zero and trends become…