Ollama

Overview

Ollama is a popular way to run open-weight language models. You can download a model and run it on your own computer, or call larger cloud-hosted models through the same app, command line and API. Ollama says prompts are never stored or used for training. It is available for macOS, Windows and Linux.

Main features

• Local models: download and run open models on your own hardware, with no usage limits on any plan.
• Cloud models: run models too large for your machine on Ollama’s servers, billed per token or from a plan’s monthly credits.
• One interface: the desktop app, the ollama command and a local API work the same for local and cloud models, so other tools can connect to it.
• Tool calling: cloud models that support tools are tested for tool calling.
• Scale: Ollama reports more than 9 million installs a month and over 1 billion model downloads.

Pricing

• Pay as you go: $0; local use is free and unlimited, cloud models are billed per token, 3 concurrent requests.
• Pro: $20 per month or $200 a year, with $60 of cloud usage a month, larger pro models and 3 concurrent requests.
• Max: $100 per month, with $300 of cloud usage a month, 10 concurrent requests and early access to new models.
• Team (early access): $500 per month, $1,000 of shared usage, unlimited users and central billing.
• Enterprise: custom pricing. Included usage resets monthly and does not roll over.

Who it is for

Developers and privacy-minded users who want to run models offline, test open models before deploying them, or connect coding and chat tools to a local model.

How to start

On macOS or Linux run curl -fsSL https://ollama.com/install.sh | sh; on Windows run irm https://ollama.com/install.ps1 | iex in PowerShell, or use the installer. Then pull a small model that fits your memory and try it in the app or the terminal.

Things to note

Local speed depends on your hardware: large models are slow without a strong GPU and enough memory. Check each model’s licence before commercial use.

Similar tools

LM Studio, Jan, GPT4All, llama.cpp, Open WebUI.