Skip to content
AIupick
Run AI locallyby Ollama Inc.

Ollama

Downloads and runs open models such as Gemma, Llama and Qwen on your computer, with a local API and optional cloud models.

Free plan
Verified free plan
Paid from
Pro $20.00 / month
Known for
Cloud models, ollama launch, Apps
Facts last checked
2026-10-11
Visit official site (opens in a new tab)

What makes Ollama different

Command line, app and local API for running open models

  1. 01

    Cloud models

    Runs open models on Ollama's servers instead of your computer, through the same app, CLI or API.

    Use it for: Using big models on a computer without a strong graphics card.

    How to start

    1. Download Ollama and open the app, or run ollama in a terminal
    2. Follow the setup prompts and sign in
    3. Pick a cloud model, for example gemma4:cloud
    No plan limit stateddocs.ollama.comchecked 2026-10-11

  2. 02

    ollama launch

    Starts a coding agent such as Claude Code, Codex or OpenCode set up to use an Ollama model.

    Use it for: Coding with an agent on open models from your project folder.

    How to start

    1. Open a terminal in your project directory
    2. Run ollama launch claude (or codex, opencode)
    3. Choose a model when prompted
    No plan limit statedgithub.comchecked 2026-10-11

  3. 03

    Apps

    On macOS, connects Claude Desktop or ChatGPT Desktop so they can use Ollama models.

    Use it for: Using open models inside the Claude or ChatGPT desktop apps.

    How to start

    1. On macOS, open Ollama and select Apps
    2. Connect Claude or ChatGPT (Desktop) and follow the prompts
    3. Choose your Ollama models in Settings → Apps
    No plan limit stateddocs.ollama.comchecked 2026-10-11

  4. 04

    Ollama app

    Desktop chat window for macOS and Windows; you can drag in text files or PDFs to ask about them.

    Use it for: Chatting with a local model about a document without the terminal.

    How to start

    1. Download Ollama for macOS or Windows
    2. Open the app and follow the setup prompts
    3. Choose a local model and drag a file into the chat
    No plan limit statedollama.comchecked 2026-10-11

  5. 05

    Modelfile

    A text file naming a base model, parameters and system message, used to create your own model variant.

    Use it for: Saving a model with your own instructions and settings.

    How to start

    1. Save the instructions as a file (e.g. Modelfile)
    2. Run ollama create choose-a-model-name -f ./Modelfile
    3. Run ollama run choose-a-model-name
    No plan limit stateddocs.ollama.comchecked 2026-10-11

Privacy and your data

What happens to what you type, upload or make, from the provider's privacy and terms pages. Business plans often differ; the plan is named under each answer.

Use at workWhether the app may be used at work or in a company, as its licence or terms state.
  • The MIT licence allows anyone to use, copy, modify and sell the software without restriction, which covers use at work.github.comchecked 2026-10-11

Cost

Free planA free plan that is not time-limited. Trials are listed separately.
  • Supportedollama.comchecked 2026-10-11

    Free plan

Plans

  • Free

    Free plan

    • $0.00 / month

  • Pro

    Paid, individual

    • $20.00 / month
    • $200.00 / year, annual commitment

  • Max

    Paid, individual

    • $100.00 / month

  • Team

    Team

    • $500.00 / monthEarly access.

    Minimum seats: not confirmed, so no total price is calculated.

  • Enterprise

    Enterprise

    Amount not confirmed. See official pricing
All features7 questions, each with its source and the date we checked it

AI models

AI modelsThe AI models the official pages name for this product. Line-ups change often, so check the date.
  • Gemma, Llama, Qwen, DeepSeek, gpt-oss, Mistralgithub.com +5checked 2026-10-11

Input and output

Chat windowComes with a desktop chat window, not only a command line.
  • Supportedollama.comchecked 2026-10-11

    Plan not specified

Workflow

Local APIServes models through an API on your computer that other apps can use.
  • Supporteddocs.ollama.comchecked 2026-10-11

    Plan not specified

Requirements

Operating systemsmacOS, Windows or Linux, as officially supported.
  • macOS, Windows, Linuxdocs.ollama.comchecked 2026-10-11
Open sourceThe tool's own code is published under an open-source licence.
  • Supportedgithub.comchecked 2026-10-11

    Plan not specified

Works offlineRuns without an internet connection once a model is downloaded.
  • Not yet confirmed
HardwareWhat computer it needs or supports, such as memory or graphics cards, as officially stated.
  • Macs need macOS 14 Sonoma or newer; Apple M-series chips use CPU and GPU, Intel Macs run on the CPU only. On Windows it uses NVIDIA and AMD Radeon graphics cards. NVIDIA graphics cards need compute capability 5.0 or higher and driver 550 or newer.docs.ollama.com +2checked 2026-10-11
Esc
↑ ↓ to moveEnter to open