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 new report from the Center for Democracy & Technology identifies dark patterns in AI chatbots that maximize usage and data collection, including deceptive design choices and misleading claims about capabilities and emo…
General-purpose large language models GPT-5.2, Gemini 3.1 Pro, and Claude Opus 4.6 outperformed specialized clinical AI tools OpenEvidence and UpToDate Expert AI on medical knowledge tests, expert alignment, and real cli…
The US Commerce Department ordered Anthropic to suspend access to its Claude Fable 5 and Mythos 5 models just three days after launch, citing a jailbreak that allowed code vulnerability scanning. Anthropic argued the tec…
A developer built a Node.js blueprint using the Fablize Claude Opus agent plugin to enforce procedural integrity in multi-tool AI workflows, reducing skipped verification steps by over 95%. The system requires agents to …
A developer operating autonomous agent systems in production reveals that LLM API costs represent only 15-25% of total system expenses, with infrastructure, engineer time, and silent costs making up the rest. Real-world …
Anthropic will retire Claude Sonnet 4 and Opus 4 model IDs on June 15, requiring users to migrate to newer versions. Claude Sonnet 4.8 launches June 16-18 with dynamic workflows and effort control, features previously in…
A developer demystifies AI agents, explaining that an agent is fundamentally a model running a loop of goal, think, act, observe, and repeat, with tools and autonomy to take multiple steps. The key to building effective …
ChatPlayground, a platform bundling over 20 AI models including GPT-4o, Claude Sonnet 4, and Gemini 1.5 Flash, is offering a lifetime subscription for $59.97, down from $619. The deal replaces separate monthly fees for C…
A developer explains why AI agents often ignore explicit rules, using a React Native project example where a rule against using 'pnpm add' for native packages was broken, causing a build failure. The developer argues tha…
A developer argues that improving AI model outputs depends more on engineering the context provided to the model than on refining prompts. The developer explains that context—including system setup, documents, examples, …
A developer proposes the API notebook pattern: keeping explanation, runnable requests, expected responses, assertions, and workflow notes in small files inside the repository. This single artifact serves documentation, t…
A developer's guide explains how to build a search autocomplete system using a Trie data structure with top-K caching to meet strict latency requirements. The system must handle billions of queries per day with sub-100ms…
The U.S. Commerce Department ordered Anthropic to suspend non-U.S. citizens' access to its latest AI models, exposing Europe's heavy reliance on American AI technology. Anthropic complied by blocking all users, including…
A developer argues that current AI memory systems are fundamentally flawed because they operate like traditional search engines rather than true memory. The post highlights that AI models retrieve information based on ke…
Developer Raine released consult-llm, an open-source tool that lets AI agents query other models for second opinions during coding tasks. The tool supports multiple providers including GPT-5.5, Gemini 3.1 Pro, and Claude…
A developer built Tiny Civilization, a browser simulation where 2-8 AI agents with distinct personalities live on an island, engaging in gathering, building, trading, stealing, gossiping, and forming grudges and alliance…
A developer implemented the SDAR gate, a gated distillation mechanism for reinforcement learning with language models, in PyTorch. The gate uses a sigmoid function to weight a per-token KL divergence loss based on the co…
Reco's security research team built an AI-powered agent that autonomously discovered and exploited high-severity vulnerabilities in Salesforce Experience Cloud sites belonging to major technology companies, extracting re…
Developer tools startup graphCTX claims its local memory system for AI coding agents achieves state-of-the-art performance, retrieving repo facts in ~1ms with flat latency even at 5,000 facts, outperforming cloud-based a…
Open-weight local language models in mid-2026 have nearly matched frontier performance for everyday tasks, driven by engineering advances in sparse attention and mixture-of-experts architectures. DeepSeek's V4 and other …