Management Commands

The CLI you'll actually use

Ollama's CLI is small — under a dozen commands you'll touch regularly. Knowing them cold removes friction from every other lesson in the quest.

Command	What it does
`ollama list`	List installed models with disk size
`ollama show NAME`	Show metadata: parameters, template, capabilities
`ollama show NAME --modelfile`	Print the Modelfile that produced the model
`ollama ps`	List currently loaded models with VRAM/RAM usage
`ollama stop NAME`	Unload a model from memory now
`ollama cp SRC DST`	Copy a model (alias). Cheap because blobs are shared
`ollama rm NAME`	Remove a model (frees disk)
`ollama pull NAME`	Download / update a model
`ollama run NAME`	Interactive chat REPL (pulls if missing)
`ollama create NAME -f Modelfile`	Build a derived model from a Modelfile
`ollama serve`	Run the daemon manually (rarely needed)

The two commands you'll actually live in

ollama ps is the canary. It tells you which models are loaded, how much memory each is using, and when each will be evicted (the UNTIL column). When something feels slow, that's the first place to look. ollama show NAME is the second — you'll use it to confirm capabilities (tools, vision, thinking) before debugging why something doesn't work.

Code

Daily driver flow·bash

# What's installed?
ollama list

# What's currently loaded in memory?
ollama ps

# What does this model actually support?
ollama show qwen2.5:7b
# Look for: capabilities (tools / vision / thinking), template, parameter_size, quantization

# Stop a model now (free memory)
ollama stop qwen2.5:7b

# Force-load a model into memory without prompting
curl http://localhost:11434/api/generate -d '{"model":"qwen2.5:7b","keep_alive":"30m"}'

# Force-unload a model
curl http://localhost:11434/api/generate -d '{"model":"qwen2.5:7b","keep_alive":0}'

Disk and version hygiene·bash

# Disk usage by Ollama
du -sh ~/.ollama/models 2>/dev/null

# What versions of each family do I have?
ollama list | awk '{print $1}' | sort

# Delete an old quant once you've moved to a new one
ollama rm llama3.1:7b-instruct-q4_0

# Update a model to the latest tag
ollama pull qwen2.5:7b

The CLI you'll actually use

The two commands you'll actually live in

Code

External links

Exercise

Progress

Comments 0