Auto-detect GPU and memory to find which local LLMs fit your VRAM, context window and quantization. Calculations stay local.
