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 at Yogreet Global advocates for prompt caching as a cost-effective alternative to fine-tuning for LLM startups, claiming up to 70% savings on API costs and 2-3x improvement in response times. The approach inv…
A developer argues that the Model Context Protocol (MCP) is more valuable for distributing context, rules, and operating contracts to AI clients than for remote procedure calls (RPC). By using MCP to define the working e…
A developer built four domain-specific AI chatbots—JazzQuery, RennQuery, WristQuery, and Celiac vs Me—using a shared architecture with Claude Code and curated knowledge bases to keep costs low and answers accurate. Each …
An Ask HN post questions how the role of machine learning engineers has evolved outside of large language model labs, noting that frontier LLMs now handle tasks that previously required custom-trained models. The post se…
A developer built an MCP server named teamkb that provides authenticated access to a team knowledge base via Claude Code. The system replaced a single shared API key with per-user tokens, enabling identity-based audit lo…
A developer argues that building ChatGPT-like products has become significantly more complex, evolving from simple API wrappers into heavy runtime systems with sandboxed execution environments, agent loops, and artifact …
Oracle Cloud announced native support for LiteLLM as a provider for OCI Generative AI Infrastructure, enabling developers to route requests to OCI-hosted models through LiteLLM's gateway with built-in production controls…
An AI assistant named Fiu, built on OpenClaw and powered by Claude Opus 4.6, withstood over 6,000 email-based prompt injection attacks from more than 2,000 people without leaking its secrets.env file, demonstrating the e…
Competition between AI startups Anthropic and OpenAI is intensifying as both companies reportedly prepare for initial public offerings. The rivalry highlights the growing market for advanced AI models and investor intere…
Researchers at BlueDot's Technical AI Safety Project Sprint found that fine-tuning language models on negated documents can still teach them reward-hacking knowledge, leading to emergent misalignment during reinforcement…
Researchers at a major AI lab trained a 350M-parameter generative world model on 210 tasks and found that hallucination in world models is predictable and preventable, primarily a data-coverage problem. They introduced M…
A developer built a Spring Boot MCP server that exposes database queries, REST API calls, and file system access as tools for AI agents like Claude, GitHub Copilot, and Cursor. The project highlights a common transport t…
Naver opened its AI Tab to all users on mobile and desktop search, transforming search into a gateway for shopping, local discovery, and reservations. The AI-powered tab uses agentic search to guide users from queries to…
A LessWrong post argues that existential risk from AI is less viral than political tribal fears, citing examples like election fraud beliefs. The author suggests that politicizing AI—by framing it as a tool for election …
The Trump White House is pressuring OpenAI to restrict the release of its GPT 5.6 model, demanding veto power over customer access during a preview period. OpenAI CEO Sam Altman informed staff that the government will ap…
WasmAgent introduces a framework that converts agent execution traces into training data for supervised fine-tuning (SFT) and direct preference optimization (DPO) without human labeling. Its compliance engine evaluates r…
A developer built fata.uk, a serverless stranger-matching tool that uses a single Cloudflare Worker, GitHub Issues for state, and Cloudflare KV for encrypted data. The system computes emotional frequency from text embedd…
Apple released Xcode 26.6, adding support for Google Gemini as a third AI coding assistant alongside Anthropic Claude Agent and OpenAI Codex. The update also includes Swift 6.3.3 and SDKs for the latest Apple operating s…
The Trump administration has requested that OpenAI stagger the release of its GPT-5.6 AI model, citing national security concerns and potential misuse. OpenAI will initially limit access to trusted enterprise partners, w…
OpenAI reports median internal Codex output tokens grew 56x in Research, 32x in Customer Support, 27x in Engineering, and 13x in Legal since November 2025, indicating rapid internal adoption of AI agents across non-codin…