Ollama

Ollama is a free, open-source tool for downloading and running open AI models such as Gemma, Qwen, gpt-oss and DeepSeek on your own Mac, Windows or Linux computer. Its optional cloud models are pay per token, with Pro at $20 a month and Max at $100, as of Nov 6, 2026.

2:41 Ollama is a free, open-source program (MIT license) that downloads open AI models and runs them on your own Mac, Windows or Linux computer with one command, such as ollama run gemma4. It has a chat app for macOS and Windows, a command line, and a local API that other apps (including Claude Code and Codex) can use instead of a cloud service. Running models on your own machine costs nothing; Ollama also offers larger cloud models, billed per token, with Pro at $20 a month and Max at $100 (as of Nov 6, 2026).

Pick Ollama if you are comfortable with a terminal, want a local model behind other tools, or want the shortest path from nothing to a running model. Pick LM Studio if you prefer a full desktop app for browsing models and tuning settings without commands. For the biggest open models that will not fit your computer, a hosted service such as OpenRouter or Hugging Face may be simpler. See Run AI Locally.

Ollama's logo, a black outline drawing of a llama's head on white Ollama

Quick facts

Maker	Ollama
Get it	ollama.com/download: macOS, Windows, Linux
Price	Free locally; cloud pay as you go, Pro $20 a month, Max $100, Team $500 (as of Nov 6, 2026)
License	MIT (open source)
Models	Gemma 4, gpt-oss, Llama, Qwen, DeepSeek, Mistral and many more in the library
API	Local at port 11434, OpenAI compatible
Best for	Running open models locally, and as a back end for coding agents

How to install Ollama

0:33 On macOS or Linux, run curl -fsSL https://ollama.com/install.sh | sh; on Windows, in PowerShell, irm https://ollama.com/install.ps1 | iex. Or download the installer (Ollama.dmg or OllamaSetup.exe) from Ollama's download page. Then ollama run gemma4 downloads Google's Gemma 4 and opens a chat in the terminal. Ollama on Apple Silicon Macs runs on Apple's MLX since a Mar 30, 2026 preview, which Ollama says is its fastest yet there.

Ollama commands

Command	What it does
`ollama run <model>`	Download if needed and chat with a model, or answer one prompt
`ollama pull <model>`	Download a model
`ollama ls`	List downloaded models
`ollama ps`	Show models loaded in memory
`ollama stop <model>`	Unload a running model
`ollama rm <model>`	Delete a model
`ollama create -f Modelfile`	Build a custom model (system prompt, settings)
`ollama launch`	Set up a coding tool (Claude Code, Codex, OpenCode, Droid, VS Code) to use Ollama
`ollama serve`	Start the server without the app

Source: Ollama CLI reference, read Nov 6, 2026.

Features

3:39 The desktop app. Since Jul 30, 2025, Ollama on macOS and Windows has a chat window: pick a model, drop in a PDF, text file or image, and raise the context length in settings for long documents (it uses more memory).

5:05 A local API for other apps. Ollama serves models at http://localhost:11434, with an OpenAI-compatible API at /v1, so tools written for OpenAI can point at your own computer. Since Jan 2026 it also speaks Anthropic's API, so Claude Code can run on local models, and ollama launch (Ollama 0.15 or later) sets up Claude Code, Codex, OpenCode and Droid in one step.

The model library. Ollama packages a large catalog of open models, each in several sizes and quantizations; for example Gemma 4 from a 4.6 GB download, gpt-oss 20b at 14 GB, and Llama 4 Scout at 67 GB. Recent additions on Ollama's blog include Meta's Muse Glimmer (Aug 10, 2026), Nvidia Nemotron models, speech models (Oct 24, 2026) and experimental image generation (Jan 20, 2026).

1:49 Cloud models. Models too big for a home computer (Ollama lists DeepSeek V4.1 Flash, GLM-5.3 and Kimi K3.1 Pro among them) run on Ollama's servers with the same commands, tagged cloud. Ollama says its cloud keeps zero data, is hosted in the US and Europe, and never trains on your data (Aug 31, 2026 post on its pricing).

Plans and prices

Plan	Price	What you get
Local	Free	Any model on your own hardware, no account needed
Pay as you go	$0 plus usage	Cloud models per token, 3 requests at once
Pro	$20 a month or $200 a year	$60 of cloud usage a month, larger Pro models, 3 requests at once
Max	$100 a month	$300 of usage a month, 10 requests at once, early access to new models
Team	$500 a month (early access)	Unlimited users, $1,000 of shared usage, admin and priority support

Prices as of Nov 6, 2026. Since Aug 31, 2026, paid plans use per-token rates with a monthly credit pool that does not roll over; earlier subscribers keep their old terms until they change plans.

Ollama vs the alternatives

LM Studio: a polished desktop app with a model browser and many settings; easier for people who avoid the terminal, and also free.
llama.cpp: the open-source engine underneath many local tools; more control, more setup. See Run AI Locally.
OpenRouter: one API for hundreds of hosted models; better when you need models far bigger than your computer can hold.
Hugging Face: where the original model weights live, with hosted inference too.

Frequently asked questions

Is Ollama free? Yes. Ollama is free, open-source software (MIT license), and running models on your own computer costs nothing; only its optional cloud models are paid, from pay as you go to Pro at $20 a month, as of Nov 6, 2026.

How do I install Ollama? Download Ollama from ollama.com/download for macOS or Windows, or on macOS and Linux run curl -fsSL https://ollama.com/install.sh | sh; then ollama run gemma4 starts a model.

Which Ollama model should I use? On a computer with 8 to 16 GB of memory, Ollama's Gemma 4 E4B or gpt-oss 20b are good starts; with a 24 GB graphics card or a 32 GB Mac, Gemma 4 31B or a mid-size Qwen model. See Run AI Locally.

Is Ollama private? Models you run locally in Ollama never leave your computer; for its cloud models, Ollama says it keeps no prompts and never trains on your data.

Does Ollama have a GUI? Yes. Since Jul 30, 2025 Ollama has a chat app for macOS and Windows; the app is for those two systems, and LM Studio is the fuller desktop app.

Sources

Ollama, GitHub, read Nov 6, 2026
Download Ollama, Ollama, read Nov 6, 2026
Pricing, Ollama, read Nov 6, 2026
CLI reference, Ollama docs, read Nov 6, 2026
OpenAI compatibility, Ollama docs, read Nov 6, 2026
Ollama's new app, Ollama, Jul 30, 2025
ollama launch, Ollama, Jan 23, 2026
Blog, Ollama, read Nov 6, 2026
gemma4, gpt-oss and llama4, Ollama library, read Nov 6, 2026
How to Use Ollama (Free Local AI), Kevin Stratvert, Nov 5, 2026 (video)
What is Ollama? Running Local LLMs Made Simple, IBM Technology, Apr 8, 2025 (video)

Moments

How to Use Ollama (Free Local AI)

Kevin Stratvert

2:41 Ollama is a free, open-source program (MIT license) that downloads open AI models and runs them on your own Mac…
0:33 On macOS or Linux, run curl -fsSL https://ollama.com/install.sh | sh; on Windows, in PowerShell, irm…
3:39 The desktop app. Since Jul 30, 2025, Ollama on macOS and Windows has a chat window: pick a model, drop in a PDF, text…
1:49 Cloud models. Models too big for a home computer (Ollama lists DeepSeek V4.1 Flash, GLM-5.3 and Kimi K3.1 Pro among…