Default grounded-answering reader for the baa.ai Paddock console. 9B parameters, 8/4-bit mixed quantization, GGUF (llama.cpp — Metal, CUDA, or CPU). ~6.8 GB; recommend >=16 GB RAM (CPU) or any >=8 GB GPU. Context: run with n_ctx 4096-8192. Downloaded automatically by the Paddock installer and container; verify against SHA256SUMS.