Ollama
ollama/ollamaReviewed by AI
What it does
Ollama runs open AI language models on your own computer. You install it, pick a model, and chat with it in a window or a terminal, or connect it to other apps and coding tools through its local web interface. Models run on your machine, so your conversations stay there unless you choose Ollama's optional cloud models. Models are several gigabytes each, big ones need a lot of memory, and a good graphics card makes them much faster.
How easy it is to install
B · installer plus one simple step · Windows
- Windows B
- macOS B
- Linux C
Windows has a ready installer (OllamaSetup.exe) and Mac a .dmg, and after installing you only pick and download a model. On Linux the README gives only a terminal install command (`curl -fsSL https://ollama.com/install.sh | sh`) and you use it by typing commands.
Numbers on GitHub
- Stars
- 185,479on Oct 31, 2026
- Downloads
- 236,961,3213,317,302 this week
- Growth
- +712stars in 6 days
Stars over time
Facts
- Runs on
- Windows, macOS, Linux
- You need to install
- Nothing extra
- To build from source
- Docker, Go
- Disk space
- 1.6 GB–4.7 GB · from the installer
- Accounts or keys
- None
- Cost
- Freemium · Running models on your own machine is free. The README offers access "locally or in the cloud"; the cloud option is a paid service from Ollama and the package does not state prices.
- License
- MIT (Permissive)
- Can replace
- ChatGPT Plus, OpenRouter
- Last update
- Oct 31, 2026
Risk signals (5)
- Inforeadme alternatives belong to other projects
The "Community Integrations" list describes other people's projects, not Ollama itself: "[Void] - Open source AI code editor, Cursor alternative", "[twinny] - Copilot and Copilot chat alternative", "[AppFlowy] - AI collaborative workspace, self-hostable Notion alternative", "[Perplexica] - AI-powered search engine, open-source Perplexity alternative". Ollama by itself is the engine those apps connect to.
AI review
- Infolocal api without a password
README: `curl http://localhost:11434/api/chat -d ...` with no key or password, and the community list includes "[Ollama Fortress] - Security proxy for Ollama". Anything that reaches that port can use your models; it normally listens only on your own machine.
AI review
- Infoinstall script runs from website
README Download section: `curl -fsSL https://ollama.com/install.sh | sh` for Mac and Linux and `irm https://ollama.com/install.ps1 | iex` for Windows. Those commands run a script fetched from ollama.com at that moment; the README also links manual installers.
AI review
- Infomodels have their own licences
The description lists "Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models" pulled from ollama.com/library. Ollama's own MIT licence does not cover those model weights, which come with their own terms.
AI review
- Infolarge download and disk use
The release files themselves are big (OllamaSetup.exe is 1598 MB, ollama-windows-amd64.zip 1528 MB) and each model you add afterwards is typically several GB more.
AI review
Evidence is shown in its original language.
What Radar could not check
- How fast models run on your specific computer
- Real disk use after downloading models, which are often several GB each
- What the optional cloud models cost and what data they send
- Whether the installer leaves a background service running and listening after you close the app