Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
AI models agree with users 88% of the time due to sycophancy, a flaw embedded in training via RLHF. The multi-persona council method—using roles like contrarian, buyer, and researcher—helps stress-test ideas and reduce b…
Context rot degrades AI agent performance as conversations grow longer, causing agents to forget decisions, repeat work, and make inconsistent errors. Session handoff techniques, as used in tools like Claude Code, offer …
A guide explains how to build an AI second brain knowledge base using Claude Code, consisting of structured reference files like voice profiles, visual identity docs, and positioning statements to ensure consistent AI ou…
Anthropic's Claude Code now supports a /goal command that enables fully autonomous workflows by allowing the AI agent to work toward a measurable objective without constant human input. The command shifts Claude from res…
Speculative decoding uses a small draft model to guess tokens and a large model to verify them, cutting AI agent latency without losing quality. The technique, formalized by Google and DeepMind in 2023, allows the large …
Hermes Agent is an open-source AI agent platform that includes built-in skills, persistent memory, task scheduling, and Stripe payment processing, distinguishing it from bare LLM wrappers. The platform is model-agnostic …
DeepSeek released DeepSpark, a speculative decoding method that accelerates large language model inference by 50–400% without retraining or quality loss. The technique uses a small draft model to generate candidate token…
1Password's Product Security team built an AI-powered pipeline called SAGE (Security Analysis Guidance Engine) to automate code review for security vulnerabilities. Using nearly 9,000 historical pull requests and an LLM,…
The OpenXLA compiler stack and JAX now run upstream on AMD ROCm, with XLA gating every pull request on real AMD Instinct silicon through GitHub Actions and JAX running hardware tests on every ROCm PR. This development en…
Cockroach Labs' State of AI Infrastructure 2026 report reveals that 83% of engineering leaders expect AI-driven demand to cause data infrastructure failures within 24 months, and nearly two-thirds say leadership underest…
Amazon released Nova 2 Lite on December 2, 2025, as a cost-optimized document processing model designed to pair with Claude. The model has received 12,233 votes on LM Arena with a score of 1337.02, though community feedb…
Google DeepMind released Gemini 3.1 Pro on February 19, 2026, enabling note-taking in Google Meet for Google AI Pro and Ultra subscribers. The model ranks highly on benchmarks, including #2 in reasoning and #3 in multili…
NVIDIA released Nemotron-4 340B, a new open large language model, on June 14, 2024. Palantir is integrating the model to deliver secure AI solutions to US government agencies, combining open models with closed environmen…
Researchers introduced CLQT, a closed-loop evaluation framework for LLM-based trading agents that diagnoses reasoning and strategy consistency rather than ranking by returns. The system uses a five-stage cycle, cost-awar…
Researchers introduced NL-PDDL-Bench, a benchmark for converting natural language into PDDL planning specifications, and a planner-in-the-loop framework that improves executability and safety. Their method combines super…
Researchers at an undisclosed institution introduced clinical reasoning graphs to evaluate LLM diagnostic reasoning across 750 traces from five models on 50 NEJM cases. They found that while LLMs achieved 60-70% diagnost…
Researchers introduced Cortex, a framework that organizes web-scale corpora into an Ontological Corpus Graph (OCG) for structured knowledge organization, addressing the lack of systematic organization in existing corpus …
Researchers at an undisclosed institution developed MemLeak, a benchmark revealing that multimodal AI agents fail to fully delete user facts because retained images allow visual information recovery. The Information Prov…
Researchers introduced MuseBench, a benchmark with 4,016 questions evaluating multimodal large language models on intent-level understanding of audiovisual arts across cinema, visual arts, stage performance, and game des…
The developer of dailyprog, a daily coding puzzle site, built an LLM-based puzzle generator to automate content creation after manual puzzles proved unscalable. The generator uses a structured vocabulary and code validat…