Assembles a new model called mymodel from the instructions in Modelfile, which typically starts with FROM llama3.2 and adds a custom system prompt, temperature, or template. The file lives in the current directory unless -f points elsewhere. Tweak and recreate the same name to iterate, and `ollama show mymodel --modelfile` to inspect the result.
Looking for more? Search all 7,657 commands — works offline, in English or Spanish, and fixes typos.