Setting up OpenCode with Ollama and sbx on Mac

The guide shows how to run OpenCode on macOS using Ollama-provided LLMs inside Docker Sandboxes (sbx). After installing sbx, OpenCode, and Ollama, it pulls Qwen 3.8 27B-mxfp8 and Gemma 4 31B-mxfp8 models, configures a sbx-kit with JSON/YAML settings that point to the host’s Ollama API, and launches OpenCode with `sbx run opencode --kit ./sbx-kit/`.