Homelab log 9: Up and running

Finally got OpenCode to run on two GPUs at one, and added a queue to help manage them more effectively. Decided to have one gpu dedicated to opencode work, while the other is for experimenting with llm based projects (currently a pokemon emerald harness).

Qwen 3.8 27b is genuinely blowing me away. It's like having opus from may 2026 running locally, for free, 24/7