OpenCode and Ollama: Qwen 2.5 Coder setup on Windows

In this Spanish-language tutorial, Hugo Paredes walks through a Windows setup that connects OpenCode to Ollama with Qwen 2.5 Coder 7B as a local coding assistant. Ollama and the model are already installed when the walkthrough begins. His reference machine has an Nvidia RTX 3070 with 8 GB of VRAM; he describes the 7B model as a balanced choice for that hardware and suggests larger variants for GPUs with more memory. These are his recommendations, rather than tested minimum requirements.

The installation uses an administrator PowerShell window and npm. Paredes covers script execution settings and discusses an installation script approval message before moving out of System32. He then edits OpenCode's JSON configuration in Visual Studio Code to connect it to Ollama and register the model. A separate folder on the Windows desktop holds the working files.

Inside OpenCode, he selects the local Ollama connection and Qwen 2.5 Coder 7B, then checks the connection with a greeting. The next prompt asks the AI agent to create a Windows script named monitor that reports current CPU and RAM usage through native system commands. The supplied description identifies the output as a .bat file, and Paredes reports generation in about 10.1 seconds. The transcript does not show a subsequent execution test of that script. This demonstrates local inference configuration rather than an independent audit of complete isolation.