AI practitioner specializing in large-model evaluation and on-premise deployment, agents, AI programming, Vibe Coding, general AI, and broader tech trends, with daily original technical articles.
This guide shows how OpenAI has embedded Codex into the ChatGPT mobile app, turning your phone into a remote control panel for a desktop‑running Codex instance, and walks through the free setup, Ollama installation, model switching, and connection workflow.
The open‑source Academic Research Skills (ARS) suite builds on Claude Code to automate the entire research‑to‑publication workflow, offering human‑in‑the‑loop quality gates, style calibration, citation checks, and a low token cost of $4‑6 per 15k‑word paper, making it especially useful for graduate students and Chinese researchers aiming to publish in English.
OpenAI has open‑sourced the openai‑realtime‑meeting‑assistant project, a Go‑based demo that lets multiple users control a Kanban board via voice using Pion WebRTC, Opus, WebSocket and the OpenAI Realtime API with function calling, and provides full setup and usage instructions.
The article explains how to enable Multi‑Token Prediction (MTP) in Qwen3.6 using a specific llama.cpp PR, achieving up to 1.5× faster local inference, details compilation steps, optimal parameters, memory requirements, and how to integrate the accelerated model with Claude Code while avoiding common pitfalls.
The article introduces a GitHub repository offering 32 ready‑to‑use HTML/CSS slide templates designed for AI agents, explains a six‑step workflow for selecting and customizing templates, evaluates the strengths and limitations of the approach, and argues that HTML is re‑emerging as a universal presentation format.
Anthropic released an open‑source Claude‑for‑Legal repository that bundles twelve role‑specific plugins, five managed‑agent recipes, and over twenty connectors, enabling law firms to automate contract review, NDA triage, amendment tracking, and other routine tasks directly within Claude’s workflow.
vLLM tops the Artificial Analysis ranking by delivering the highest throughput for DeepSeek V3.2, Qwen 3.5 397B, and MiniMax‑M2.5 on identical NVIDIA Blackwell Ultra hardware, thanks to extensive kernel‑fusion optimizations that remain in the main branch.
Unsloth adds MTP to Qwen3.6‑27B and 35B‑A3B models, delivering 1.5‑2× decoding speed gains on consumer‑grade GPUs, with ~80% draft acceptance, while providing installation steps, usage parameters, benchmark results, and guidance on suitable scenarios.
The article dissects Perplexity’s internal best‑practice guide for building Agent Skills, showing how Skill design flips conventional coding wisdom, introduces a three‑tier context‑cost model, and provides a step‑by‑step workflow, maintenance tips, and real‑world examples.
The article introduces the open‑source Qwen3.6‑35B‑A3B model, explains its MoE architecture, three‑stage LoRA fine‑tuning, shows benchmark results where it achieves 161.9 tok/s on an RTX 5090—2.6× faster than a dense 27B counterpart—and discusses deployment tips, quantized GGUF release, and known compatibility pitfalls.
