Run OpenCode with local Ollama models in Docker sandboxes on Mac
Setting up OpenCode with Ollama and sbx on Mac
A practical guide to running OpenCode with Ollama and Docker sandboxes on an M5 MacBook Pro. It covers installing the tools, pulling Qwen 3.8 and Gemma 4 models, and configuring an sbx kit to connect OpenCode to Ollama on the host. The author argues that local LLM development is now viable, with 30B models handling web app builds, and recommends sandboxing to contain potential hallucinations.
Viable local LLM development is here.
- manapause
If you have both ollama and llama.cpp installed on your MBP, and are interested in utilizing your local model as a cost-saving preprocessor for your frontier models, consider using https://github.com/Standard-Pentest/kultivait!
"kultivait init --setup" to evaluate your machine's hardware, access to frontier cli-tools, download models, and create a proxy for ollama to get started. works great with opencode!
- hkchad
Friends don't let friends use Ollama
- darthcircuit
I’ve been using OMLX on 48gb Mac. It’s a lot smoother setup than Ollama and has a built in model sizer and all that in a nice ui. Also will pre configure and launch opencode, pi, Hermes, codex and Claude out of the box with no fuss. I’ve been really happy with it.