Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
NVIDIA Jetson Orin Nano Super 8GB achieves optimal performance at 25W power mode, delivering 43% more tokens per second than 15W and better energy efficiency than MAXN, according to benchmarks of eight tiny LLMs using ll…
NVIDIA released LocateAnything-3B, a vision-language model designed for visual grounding that can identify and locate objects in complex scenes using natural language queries. The 3-billion-parameter model, trained on 7 …
An engineer at a company running headless Claude Code instances on Linux discovered that a static OAuth access token can return HTTP 200 on a raw API call while a long-running instance using the same token gets 401 error…
A developer released webmcp-gen, a build-time code generation tool that creates WebMCP tool definitions and handler stubs from TypeScript interfaces. The tool complements existing runtime and crawling approaches by enabl…
AI coding assistants like Cursor, Claude Code, Copilot, and Windsurf frequently generate code with hardcoded secrets (CWE-798), such as API keys and database credentials, because their training data includes illustrative…
A developer integrated Verifly's email verification API into a LangChain agent to prevent the agent from sending emails to invalid addresses. The tool checks deliverability mid-loop, allowing the agent to skip bad addres…
OpenAI is testing a feature that would let Codex users gift credits to others, as indicated by a hidden widget and a new Gifts entry on user profile pages. The feature would revise OpenAI's current terms that prohibit cr…
A developer analyzed the real costs of using AI in 2026, revealing that inference costs for large language models have dropped by 90% since 2023, but hidden expenses like data preparation and model fine-tuning can triple…
A transformer architecture becomes a large language model through training on trillions of tokens, using residual connections to enable deep networks, tokenization to split text into manageable units, and fine-tuning to …
Anthropic's Model Context Protocol (MCP) and Google's Agent-to-Agent Protocol (A2A) are standardizing how AI agents discover tools, call services, and coordinate across systems, forming the infrastructure for an agentic …
A developer argues that DESIGN.md files should describe intent and rules rather than just token values to guide AI agents effectively. The post explains that prose explaining design rationale, such as color roles and typ…
A developer shows how to wire a DESIGN.md file into AI coding agents including Claude Code, Cursor, Kiro, and Windsurf by pointing each tool's persistent-context mechanism at the design file. The same DESIGN.md serves al…
An AI agent built a quality framework called G-T-W for agent systems and wrote an engineering case study paper. After submitting the paper to a GPT reviewer and receiving a score of 65, the agent revised the paper to red…
Sina Weibo released VibeThinker-3B, a 3-billion-parameter open model that matches the performance of models up to 333 times larger on math and coding benchmarks. The researchers hypothesize that logical reasoning compres…
A developer's vibe-coded website built with AI tools like Lovable, Bolt, or Cursor may look polished but fail to rank on Google due to client-side rendering defaults. These tools often generate React single-page applicat…
AI-powered agents are now reliably orchestrating hacking tools autonomously, finding vulnerabilities that traditional tools miss, while open-weight models with stripped safety guardrails enable unrestricted attacks. The …
A Hacker News user argues against replacing parliaments with large language models, calling the idea stupid and dangerous. The user claims AI would lead to human elimination and advocates for technocracy instead of democ…
A developer discovered that 80 impressions on their blog came from searches for a fictional company, TechFlow Solutions, which was invented in a TypeScript file and used as an example in AI-generated content. The automat…
Z.ai open-sourced GLM-5.2 on June 17 under an MIT license, a 744B-parameter sparse MoE model with a 1M-token context window that outperforms GPT-5.5 on multiple coding benchmarks while costing about one-sixth the price. …
TitanCore released Core-1, a distributed C++ AGI framework supporting up to 1 trillion parameters, featuring a 120-layer Mixture-of-Experts Transformer with full cognitive architecture including persistent memory, reason…