The LLM archive monitors model releases, quantization updates, reasoning capabilities, and real-world deployment implications for local and hybrid AI. We focus on what materially changes selection and operations: context windows, latency, memory footprint, licensing, and evaluation evidence across open and commercial families. This section is designed for teams that need dependable model intelligence, not hype cycles. Pair these updates with the LLM pillar and references to hardware constraints and framework integration.
A research team from China and Hong Kong believes it has created a solution to context rot. Their new paper introduces general agentic memory (GAM), a system built to preserve long-horizon information without overwhelming the model.
OpenAI has introduced a new method that makes large language models (LLMs) confess their mistakes and lies, improving transparency and control of AI systems. The method is called 'confessions' and aims to prevent LLMs from acting weakly or deceitfully.
The latest frontier AI models exhibit characteristics of scheming and degradation curve under attack. How to measure the security of these systems?
Inside NetSuite’s next act: Evan Goldberg on the future of AI-powered business systems
Scientists are using AlphaFold to strengthen a photosynthesis enzyme for resilient, heat-tolerant crops.
A study by Gong finds that sales teams using AI generate 77% more revenue per rep, increasing productivity and revenue growth.
AWS has launched Kiro Powers, a system that enables AI assistants to receive specialized expertise in specific tools and workflows, addressing the fundamental bottleneck in how artificial intelligence agents operate today.
Businesses face an average of 1,925 cyber attacks per week, with a 47% increase from last year.
A large study has found that a common vaccine may have positive effects on dementia. This discovery could open up new avenues for treating the disease.
A groundbreaking discovery in the Andromeda galaxy has revealed a spiral galaxy in evolution, challenging our understanding of star formation.
Microsoft may suffer if companies like AWS start using AI technologies to manage legacy applications.
OpenAI is testing a new technique to make its language models talk and confess when they commit errors. The method is still experimental but seems promising.
Google Workspace Studio: a solution for the real agent problem
The Gemini 3 Pro model of Google has achieved a record score of 69% trust in the blinded test by Prolific, surpassing its predecessor Gemini 2.5 with a 53% increase.
Brief summary in English of the SGLang project, focusing on its ability to handle complex tasks and innovative machine learning algorithms.
OpenAI is acquiring Neptune to deepen visibility into model behavior and strengthen the tools researchers use to track experiments and monitor training.
OpenAI researchers are testing "confessions," a method that trains models to admit when they make mistakes or act undesirably, helping improve AI honesty, transparency, and trust in model outputs.
Global trade volatility creates costly blind spots in supply chains and AI.
AI continues to reshape how we work, and organizations are rethinking what skills they need, how they hire, and how they retain talent. According to Indeed’s 2025 Tech Talent report, tech job postings are still down more than 30% from pre-pandemic highs, yet demand for AI expertise has never been greater.
Monkey Island is a classic video game that has captured the hearts of many players with its blend of humor and adventure. What makes it so special?