Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Vercel's AI SDK DefaultChatTransport, built for HTTP and serverless platforms, fails in production scenarios requiring persistent sessions, causing issues like undetectable disconnections and continued billing. The SDK's…
Three recent AI events reveal the fragility of relying on third-party model APIs. Anthropic suspended access to its Fable 5 and Mythos 5 models after a US government export control directive, causing a global blackout. M…
Developer Rufus SD released Prismag, an open-source CLI tool that routes different parts of a single prompt to different AI models using @@tags, enabling per-block model selection in terminals and IDEs like Cursor and Cl…
A developer with over a year of experience using LLMs and agents for coding warns that 'vibe coding' creates a false sense of productivity. The shift from writing code to reviewing AI-generated code increases cognitive l…
Pydantic AI released on-demand capabilities that allow AI models to defer loading instructions, tools, and settings until needed, reducing token costs. The feature, requested by users, uses a `defer_loading=True` paramet…
Next.js App Router introduces a new error boundary model using `error.js` files that can be nested at any route level, allowing errors to be contained locally rather than propagated globally. A developer explains how to …
After three years of building evaluations for financial AI agents, a former hedge fund analyst reveals that standard benchmarks fail to capture the nuance required for equity research. The author developed internal evals…
A developer who built a Swift tutorial series and a mascot generator app with Google AI Studio in the same week found that each workflow builds a different kind of understanding. Writing Swift line by line sharpens mecha…
Tencent has begun internal testing of Xiaowei, an AI assistant embedded in WeChat that accepts text and voice input to execute tasks across the app's mini-program ecosystem, with a public rollout targeted for Q3 2026. Th…
Google DeepMind distinguished engineer Vladimir Feinberg advised aspiring AI researchers to "work like a dog" to land jobs at top labs like OpenAI, Anthropic, and Google DeepMind, emphasizing intent, mathematical maturit…
A developer argues that AI should be used to review large code diffs, as human reviewers should focus on providing out-of-distribution knowledge that the AI lacks, such as deprecated services or codebase-specific princip…
Baidu is pursuing full-stack AI capabilities by building its own chips, models, and cloud infrastructure while gaining access to restricted Nvidia H200 GPUs. The company's Kunlunxin chip unit could be valued at $28 billi…
Japanese AI startup Sakana AI launched Fugu, a system that orchestrates multiple language models to match or exceed the performance of Anthropic's Fable and Mythos models on benchmarks. Fugu operates as a single model vi…
Micheal Lanham released the second edition of "AI Agents in Action," a practical guide to building agentic systems covering personas, tools, memory, and multi-agent patterns. The 392-page book, published by Manning in Ju…
Z.ai's GLM-5.2, a 756B-parameter model with a 1M-token context window, competed against Claude Opus in a Tech Stackups test building a 3D platformer in raw WebGL. Claude Opus finished in 33m 30s at an estimated $21.92, w…
Microsoft Research's AutoGen framework enables multi-agent AI systems where specialized agents collaborate through structured conversations to solve complex tasks. The open-source framework, which gained popularity for i…
Tezcode, an AI software factory based in Tashkent, published a practical guide on building AI agents for businesses in 2026. The guide explains that AI agents differ from chatbots by making decisions, working with extern…
A developer traces the 70-year history of AI from symbolic systems to modern transformer-based models, highlighting key milestones like the 1956 Dartmouth Workshop, expert systems, the first AI winter, the rise of statis…
Researchers introduced Reflective Masking (RM), a post-training method that enables Mask Diffusion Models to iteratively revise their own outputs by re-masking uncertain tokens and remembering previous attempts. Tested o…
Headroom launches a context compression layer for AI agents that reduces token usage by 60–95% while preserving answer quality. The open-source tool offers multiple integration modes including a library, proxy, agent wra…