Local AI Home Lab
01 Jan 2026
• 1 min read
Self-hosted LLM inference stack on an ASRock BC-250 with a 40-CU GPU unlock.
A self-hosted AI inference stack running Ollama and llama.cpp on an ASRock BC-250 (AMD APU, 40-CU unlock, custom SMU governor), serving local models for private retrieval and automation workloads.