Over Half of Developers Now Write Less Than 20% of Their Code
JetBrains surveyed 15,000+ developers in 2026. Over half now manually write less than 20% of their code. One in five writes zero lines ...
Cursor /multitask: Parallel Agents and the Cost You’re Not Tracking
Cursor shipped /multitask, Security Review, and context breakdowns in five weeks. Here's what parallel agents ...
GLM-5.3-Flash: The MIT-Licensed Open Model Topping Coding Benchmarks
Z.ai's GLM-5.3-Flash is MIT-licensed, scores above Claude Opus 4.8 on coding benchmarks, and costs 30x ...
OpenClaw 2.0: Multiplayer AI Agents Are Here
OpenClaw 2.0 ships collaborative agent sessions — multiple users, one live agent, full context. Here's ...
Claude Fable 5.1: 75% Cheaper Cache, Stronger Agents
Anthropic released Claude Fable 5.1 yesterday with 75% cheaper cache reads and up to 45% ...
Runway Solaris: The Interface World Model That Renders UI Without Code
Runway's Solaris generates app interfaces as live video frames — no HTML, no JavaScript, no DOM. Here's what it means for developers and ...
VS Code 1.135 Rubber Duck: AI Agents Critique Each Other’s Code
VS Code 1.135 ships Rubber Duck, an experimental AI critic agent that uses a different ...
OpenCode Hits 194K Stars: The Free Claude Code Alternative
OpenCode crossed 194K GitHub stars in August 2026 — ahead of Claude Code. Here's what ...
Claude Code Compliance API: Local Session Transcripts Are Here
Anthropic added Compliance API endpoints that expose Claude Code session transcripts to enterprise teams. Here ...
IBM Granite 4.2: Free Local Reasoning Models Worth It?
IBM Granite 4.2 ships 3B, 8B, and 30B reasoning models under Apache 2.0 — run ...