GitHub has released the Copilot SDK in technical preview. It exposes the Copilot CLI agent runtime as a programmable interface and supports custom tools as well as MCP server connections.
A case study on building a real-time analytics platform with Cloudflare Workers, Hono.js, and Supabase. It covers practical edge-computing design decisions, including global P95 event ingestion of 47 ms and cookie-free session management.
Anthropic has released the mid-sized model Claude Sonnet 4.6. In Claude Code evaluations, 70% of users preferred it over Sonnet 4.5, and 59% preferred it over Opus 4.5, while pricing remains unchanged.
VectorWare has announced the first implementation of Rust's Future trait and async/await running on GPUs by adapting the Embassy executor to a GPU environment.
Alibaba open-sourced AliSQL 8.0, which combines MySQL-compatible OLTP, DuckDB-based high-speed OLAP, and HNSW vector search into a single database. This article considers what that means amid MySQL stagnation and PostgreSQL's momentum.
A roundup of four topics around AI agent development and operations in February: a study showing AGENTS.md may be counterproductive, Continue.dev's CI-integrated AI checks, AWS Strands Agents' built-in session persistence, and Docker Shell Sandbox for isolated agent execution.
Google has launched a public preview of the Developer Knowledge API and MCP Server, letting generative AI tools access official documentation for Google Cloud, Android, Firebase, and other Google technologies.
Using Jeff Geerling's article as a starting point, this piece looks at how low-quality AI-generated contributions are increasing the burden on open source maintainers, along with responses from curl and GitHub.
How to configure VRAM/main memory split on the GMKtec EVO-X2 (Strix Halo) for local LLM inference. A 29.6GB model ran fine with just 8GB of dedicated VRAM.
Building an NSFW-capable local LLM on the GMKtec EVO-X2 (Strix Halo). Getting GPU inference at ~11 tokens/s with LM Studio and MS3.2-24B-Magnum-Diamond.
An explanation of why Qwen-Image-Edit's VAE is so heavy, how HunyuanImage 2.1 chose a 32x high-compression VAE instead, and how Kohya's memory-optimization work fits in.
