Local LLM Development Setup with Ollama and Docker Sandboxes
September 10, 2026
Running 30B parameter models like Qwen 3.8 on 48GB Apple Silicon is viable using Ollama's MLX integration. To prevent hallucinations from affecting host systems, integrate OpenCode with Docker sandboxes (sbx) for containerized execution.
HOW THIS AFFECTS YOU
●
builderYou can develop with large local models in a sandboxed environment using Apple Silicon.