Ollama lets you download and run language models on your own computer, then manage them from a command line or use them through a local REST API. The core workflow is: install Ollama for your platform, run a model with ollama run, inspect or remove local models with a few CLI commands, and create a custom model when you need different settings or a local GGUF file.
Install Ollama for your platform
Use Ollama’s official quickstart to choose the current instructions for macOS, Windows, Linux, or Docker. The available method differs by platform: the quickstart links macOS and Windows downloads, provides a Linux shell installer and manual instructions, and points to the official Docker image.
As an Amazon Associate I earn from qualifying purchases.
For Linux, the quickstart shows this installer command:
curl -fsSL https://ollama.com/install.sh | shThat command is the documented Linux path, not a universal installation command. For other platforms or a container deployment, follow the corresponding current instructions in the Ollama documentation index.
Choose a model that fits your machine
Ollama’s model library contains models and variants with very different download sizes. For example, the quickstart lists Llama 3.2 1B at 1.3 GB, Llama 3.2 3B at 2.0 GB, Llama 3.1 70B at 40 GB, and Llama 3.1 405B at 231 GB. These are example artifact sizes, not estimates of the RAM needed to run a model.
As rough minimum RAM guidance, Ollama’s quickstart gives at least 8 GB for its 7B examples, 16 GB for 13B, and 32 GB for 33B. Treat these as guidance rather than a guarantee: a model’s quantization, context length, device, and workload affect whether it runs well. Check the current Ollama model library and the requirements for the exact model tag before downloading.
Download and start a model
The quickest way to begin is to run a model by name. Ollama downloads it if needed and starts an interactive session:
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →ollama run llama3.2Model tags identify variants; the quickstart examples include llama3.2:1b, llama3.1:70b, and llama3.2-vision:90b. Check the model library for currently available names and tags rather than assuming examples will remain unchanged.
To download a model without starting an interactive session, or to update a local copy, use:
ollama pull llama3.2Ollama says a pull can update a local model by downloading only the difference. Replace the example name with the model and tag you want.
Inspect, stop, copy, or remove local models
Use the command that matches what you need to check or change:
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minute| Task | Command | What it does |
|---|---|---|
| See downloaded models | ollama list |
Lists models stored locally. |
| See models currently loaded | ollama ps |
Shows models active in memory. |
| Inspect a model | ollama show llama3.2 |
Displays information about the named model. |
| Stop a running model | ollama stop llama3.2 |
Stops the named model. |
| Copy a model under another name | ollama cp llama3.2 my-llama |
Creates a copy with the destination name. |
| Remove a local model | ollama rm llama3.2 |
Deletes the named model from local storage. |
Substitute the exact model name or tag shown by ollama list when using a management command.
Import a local GGUF or customize a model
Import a GGUF file
To build a model from a local GGUF file, create a file named Modelfile whose FROM line points to that file. Then create and run the named Ollama model:
Rank #4
ollama create example -f Modelfile
ollama run exampleThe path in FROM should identify the GGUF file on your machine. Consult the current importing guide for supported formats and exact file requirements.
Set parameters or a system message
A Modelfile can also start from a model in Ollama’s library and specify parameters or a system message. For example:
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11FROM llama3.2
PARAMETER temperature 0.7
SYSTEM "You are a concise writing assistant."Save the file as Modelfile, then build and run the customized name:
ollama create my-assistant -f Modelfile
ollama run my-assistantThis illustrates a basic customization pattern, not every supported directive. The current Modelfile reference documents additional details.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Use Ollama through its local API
Ollama exposes a REST API at localhost:11434. The quickstart shows these basic POST requests; they assume the Ollama service is running locally.
Generate from a prompt
curl http://localhost:11434/api/generate -d '{
"model": "llama3.2",
"prompt": "Explain local language models in one paragraph."
}'Send a chat message
curl http://localhost:11434/api/chat -d '{
"model": "llama3.2",
"messages": [
{"role": "user", "content": "Explain local language models in one paragraph."}
]
}'When Ollama is not running through its desktop application, the quickstart documents starting the service with:
ollama serveThese examples show the basic endpoints and request shapes; see the current API reference for endpoint details and full request and response behavior.
Pick an interface based on how you work
The Ollama quickstart links a directory of community integrations, including web and desktop clients, terminal tools, and cloud deployment integrations. It is a list, not a tested ranking or endorsement. Choose by where inference will run and how you want to interact:
Quick Recap
- Terminal: Run models directly with the Ollama CLI.
- Desktop or web chat: Look for a client that connects to your local Ollama instance if you want a standalone chat interface.
- Your own application: Use the local REST API when your app needs to send prompts or chat messages programmatically.
- Cloud deployment: Check where the integration sends requests and runs inference; a cloud integration is not the same as running a model solely on your computer.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
