Chat with Gemma 4 interactively

run with ollama

// what it does

ollama run opens a REPL against the model, pulling it first if needed. Inside, /set parameter tunes sampling, /show info prints the effective modelfile, and /bye exits. The default gemma4 tag is E4B, a good balance of quality and a 5-8 GB memory footprint.

// shell

$ ollama run gemma4

// gotcha

To paste a multi-line prompt, wrap it in triple quotes (""") — otherwise Ollama submits on the first newline and treats the remaining lines as a new prompt.

// resources