Ollama Related Commands

Ollama provides a variety of command-line tools (CLI) for users to interact with locally running models.

Basic format:

ollama <command> [args]

We can useollama --helpto see what commands are available:

Large language model runner

Usage:
  ollama [flags]
  ollama [command]

Available Commands:
  serve       Start ollama
  create      Create a model from a Modelfile
  show        Show information for a model
  run         Run a model
  stop        Stop a running model
  pull        Pull a model from a registry
  push        Push a model to a registry
  list        List models
  ps          List running models
  cp          Copy a model
  rm          Remove a model
  help        Help about any command

Flags:
  -h, --help      help for ollama
  -v, --version   Show version information

1. Usage

ollama [flags]: Run ollama with flags.
ollama [command]: Run a specific ollama command.

2. Available Commands

serve: Start the ollama service.
create: Create a model from a Modelfile.
show: Display detailed information about a model.
run: Run a model.
stop: Stop a running model.
pull: Pull a model from a model registry.
push: Push a model to a model registry.
list: List all models.
ps: List all running models.
cp: Copy a model.
rm: Delete a model.
help: Get help information about any command.

3. Flags

-h, --help: Display help information for ollama.
-v, --version: Display version information.

Complete example:

Command	Description	Example
`ollama run`	Run a model. If it does not exist, it will be pulled automatically.	`ollama run llama3`
`ollama pull`	Pull a model. Download the model from the registry without running it.	`ollama pull mistral`
`ollama list`	List models. Display all locally downloaded models.	`ollama list`
`ollama rm`	Delete a model. Remove the local model to free up space.	`ollama rm llama3`
`ollama cp`	Copy a model. Copy an existing model to a new name (for testing).	`ollama cp llama3 my-model`
`ollama create`	Create a model. Create a custom model from a Modelfile (advanced).	`ollama create my-bot -f ./Modelfile`
`ollama show`	Show information. View the model's metadata, parameters, or Modelfile.	`ollama show --modelfile llama3`
`ollama ps`	View processes. Display currently running models and VRAM usage.	`ollama ps`
`ollama push`	Push a model. Upload your custom model to ollama.com.	`ollama push my-username/my-model`
`ollama serve`	Start the service. Start Ollama's API service (usually runs automatically in the background).	`ollama serve`
`ollama help`	Help. View help information for any command.	`ollama help run`

1. Pulling and Deleting Models

pull
Pull a remote model to the local machine.

ollama pull <model>

rm / remove
Delete a local model.

ollama rm <model>

list / ls
List all local models.

ollama list

2. Running Models

run
Run the model in interactive mode without exiting.

ollama run <model>

Can include system information and prompt:

ollama run <model> -s "<system>" -p "<prompt>"

run + script
Read prompt from a file:

ollama run <model> < input.txt

When you enterollama runAfter entering the chat interface, you are no longer operating the command line, but talking to the AI. At this point you can use the/shortcut commands starting with a prefix to control the conversation:

/byeor/exit：Most important!Exit the chat interface and return to the command line.
/clear: Clear the current context memory (start a new conversation).
/show info: View detailed parameter information of the current model.
/set parameter seed 123: Set a random seed (advanced technique, used to reproduce results).
/help: View all available shortcut commands in the chat.

3. Inference Interface (One-time Execution)

generate
Perform a single inference and output text.

ollama generate <model> -p "<prompt>"

4. Creating and Modifying Models

create
Create a local model from a Modelfile.

ollama create <model-name> -f Modelfile

cp
Copy a model to a new name.

ollama cp <src> <dst>

5. Server Related

serve
Start the Ollama local service (default 11434).

ollama serve

run serverless
Whenollama runIt automatically starts the background service; no need to execute it separately.

6. Model Information

show
View model metadata, parameters, and templates.

ollama show <model>

7. Special Parameters

Most of these parameters can be used in run/generate:

--num-predict <number>    限制输出 token 数
--temperature <float>     控制随机性
--top-k <int>             采样范围
--top-p <float>           核采样
--seed <int>              固定随机性
--format json             输出 JSON
--keepalive <seconds>     会话保持时间

8. Modelfile Directives

Used when building a model:

FROM <model>: Base model
SYSTEM "xxx": Set system prompt
PARAMETER key=value: Set default parameters
TEMPLATE "xxx": Customize Chat template
LICENSE "xxx": Set License
ADAPTER <file> / WEIGHTS <file>: Load LoRA or additional weights

9. API (when serve is running)

REST endpointshttp://localhost:11434/api：

/api/generate: Text generation
/api/chat: Chat streaming interface
/api/pull: Remote pull
/api/tags: Local model list

Example call (curl):

curl http://localhost:11434/api/generate \
  -d '{"model":"qwen2.5","prompt":"hello"}'

10. Advanced

Run with custom parameters:

ollama run <model> --temperature 0.2 --top-p 0.9

Persistent session (preserving context):
Sessions are automatically managed by the model's internal cache, without the need for additional commands.

Other extensions

Project practice