Ollama in VS Code: Qwen 2.5 Coder and Gemma 4 setup

In this Spanish-language tutorial, Hugo Paredes demonstrates how to connect Ollama to Visual Studio Code's chat and use a local LLM as a coding assistant. His setup uses an RTX 3070 with 8 GB of VRAM. He installs Qwen 2.5 Coder 3B alongside an existing 7B model; the demonstrated 3B download is 1.8 GB.

The setup follows VS Code's model settings: choose Add Models, select Ollama, confirm the group name, and accept localhost port 11434. Installed models then appear in the chat's model selector. The walkthrough also shows how to apply generated code to an editor and save it.

The examples use different tasks. Qwen 2.5 Coder 3B receives a PowerShell prompt to cancel any previous shutdown and schedule a forced Windows shutdown in 60 minutes. The 7B model receives a batch-script request to clear the DNS cache and ping Google's servers. Gemma 4 4B receives a restaurant landing-page prompt using HTML, CSS and JavaScript.

Paredes praises Gemma's output as better than Qwen's, but these examples use different prompts and do not establish a controlled comparison. The transcript shows code generation and saving rather than execution tests. He presents the setup as useful for basic work while reserving more demanding tasks for cloud tools such as Claude Code or Codex.