Player not loading? Watch on YouTube
Jonathan explains his AI setup on a Dell Precision 5530 running EndeavourOS, an Arch-based Linux distribution, with RAM upgraded to 24 GB. He uses Ollama to run models locally and Cherry Studio as the desktop chat interface. His limited compute keeps him on smaller models, including Qwen 3.5 with 2 billion parameters and LFM 2.5 Thinking.
The walkthrough covers Ollama installation, starting its server with ollama serve, listing models and finding model commands. In Cherry Studio, he configures the Ollama API host on port 11434, fetches the model list and selects a model for chat. Rust questions provide examples of this local LLM workflow, which he uses as an offline ChatGPT alternative. He says he does not use agentic features in this local chat setup.
For coding, Jonathan switches to OpenCode's terminal interface. He shows the command palette, model selection and a workflow that reviews a plan before switching to build mode. The demonstration uses OpenCode Zen to generate a village website with HTML, CSS and JavaScript, then previews it through VS Code's live server. Although he shows Ollama models in OpenCode's selector, this coding example uses Zen. His claim that the setup costs nothing to start applies to September 2026; he warns that tool availability and pricing may change.
