The Digital Press

Ollama has released new coding models GLM-4.6 and Qwen3-coder-480B on its cloud service, with improved integrations for popular development tools like VS Code, Zed, and Droid. The update also includes faster, more reliable tool calling for Qwen3-Coder-30B and offers local access to the largest model for users with over 300GB of VRAM.

Why it matters: Developers gain easy access to powerful AI coding assistants integrated directly into familiar coding environments.

The big picture: Cloud and local options provide flexibility for different hardware capabilities and workflows in AI-assisted programming.

The stakes: Local use of the largest model requires extremely high VRAM hardware, limiting accessibility for most users.

Commenters say: Users praise GLM-4.6’s reasoning and usability but express concerns about hardware demands and Ollama’s focus on monetizing model inference over local app support.