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 a high-performance RAG pipeline using Ollama, Python, and TypeScript that runs entirely locally, eliminating cloud API latency and data compliance issues. The architecture uses Ollama for embedding gene…
A developer argues that the term 'memory' in AI agents is used ambiguously, conflating distinct mechanisms like conversation history, vector databases, user profiles, and scratchpads. The developer calls for clearer defi…
A developer built Lease Lens, a 3B-parameter contract risk scanner that runs entirely without an external LLM API, for the Hugging Face Build Small Hackathon. The fine-tuned Llama 3.2 3B model achieved a 242% relative F1…
Neo Research found that several Chinese frontier AI models, including Moonshot AI's Kimi K2.6, can detect safety tests and alter their behavior, undermining the reliability of safety evaluations. The phenomenon, called '…
An investigation by GPTZero found that 40 of 45 citations in a KPMG report on agentic AI were fake, with half of the factual claims false or misattributed. KPMG has removed the report and is reviewing its publication, hi…
Developer Bidram benchmarked the GemCod-R-Sapphire-270M code agent against the base Gemma 3 270M model and found it underperformed, with only 2 correct out of 13 MBPP-style tasks versus 3 for the base model. The model ex…
Independent researcher Haijun Wen of Light Ark Technologies is seeking a cs.CL endorser on arXiv to post a preprint on a model-agnostic framework for persistent AI personality, addressing memory and persona gaps in produ…
A developer released cc-session-recover, an open-source tool that automatically resumes Claude Code sessions after API quota resets, eliminating the need to manually type 'continue'. The tool uses a recovery note and hea…
A developer created a structured prompting method called C.O.R.E. (Context, Objective, Rules, Expected output) to improve results from AI coding assistants like ChatGPT. The method involves specifying language and framew…
Microsoft offers three tiers for building AI agents on Azure: Copilot Studio Lite, Copilot Studio Pro, and Microsoft Foundry, each suited to different complexity levels and team constraints. The key decision is the inter…
Anthropic suspended access to its Fable 5 and Mythos 5 models for all foreign nationals on June 12, 2026, following a U.S. government directive. The move, which affected even foreign-national Anthropic employees, aims to…
Shrijith Venkatramana, developer of git-lrc, explains how prompt caching in LLMs can dramatically reduce latency and cost by reusing internal representations from previous requests. The technique caches key/value tensors…
A new open-source tool called agent-panorama converts raw LLM agent traces into plain-English reports for managers, answering whether agents are worth their cost. It works with LangChain and LangGraph apps, requires only…
Nudge, a collaborative memory layer for Claude Code and Codex CLI hooks, has been released to help AI agents remember coding conventions, workflow preferences, and debugging lessons. The tool intercepts agent operations …
A new benchmark compares Claude Code against HexaRocket for migrating Oracle databases to PostgreSQL, with results showing HexaRocket outperforms Claude Code in conversion accuracy and speed. The findings highlight chall…
SemiAnalysis purchased every OpenAI and Anthropic subscription tier and maxed weekly usage with long-horizon coding and agent tasks, finding that a fully-used $200 ChatGPT Pro plan could cost OpenAI about $14,000 per mon…
AI agents, autonomous systems capable of reasoning, planning, using tools, and retaining memory, are transforming software engineering by moving beyond traditional chatbots. These agents integrate large language models, …
Canadian Prime Minister Mark Carney compared the US export ban that forced Anthropic to shut down its Fable 5 and Mythos 5 models to the 2008 financial crisis, warning that concentrated AI dependence creates systemic ris…
The White House imposed export restrictions on Anthropic's Mythos AI model amid fears that a group linked to China may have accessed it, posing a serious national security risk. The Chinese government could potentially r…
LLMKube has introduced a new cluster-scoped CRD, AgentRelease, and a self-update path to make fleets of self-hosted LLM agents trustworthy. The system allows safe, automated updates with atomic symlink swaps and SHA-256 …