Gemma 2 Building a Private AI Terminal with Docker, llama.cpp, and OpenCode May 9, 2026 Local AI: Running Gemma 4 with llama.cpp and Docker May 8, 2026