Fetches the 3 billion parameter llama3.2 variant, a lighter model that runs comfortably on CPU-only laptops and 4-6 GB GPUs. The tag after the colon selects the size or quantization (compressed precision) of the same base model. Compare with llama3.2:1b for an even faster, dumber model.
Looking for more? Search all 7,657 commands — works offline, in English or Spanish, and fixes typos.