██████╗ █████╗ ███╗ ██╗ ██╗ ██████╗ ██╗ ██╗███╗ ██╗ █████╗ ██╗██████╗ ██╔════╝██╔══██╗████╗ ██║ ██║ ██╔══██╗██║ ██║████╗ ██║ ██╔══██╗██║╚════██╗ ██║ ███████║██╔██╗ ██║ ██║ ██████╔╝██║ ██║██╔██╗ ██║ ███████║██║ ▄███╔╝ ██║ ██╔══██║██║╚██╗██║ ██║ ██╔══██╗██║ ██║██║╚██╗██║ ██╔══██║██║ ▀▀══╝ ╚██████╗██║ ██║██║ ╚████║ ██║ ██║ ██║╚██████╔╝██║ ╚████║ ██║ ██║██║ ██╗ ╚═════╝╚═╝ ╚═╝╚═╝ ╚═══╝ ╚═╝ ╚═╝ ╚═╝ ╚═════╝ ╚═╝ ╚═══╝ ╚═╝ ╚═╝╚═╝ ╚═╝
Find out which AI models your PC can run
We detect your hardware automatically and show the best models for you.
Recommended Models
116 models
Gemma 3 12BChat
Llama 3.2 Vision 11BMultimodal
Mistral Nemo 12BChatGemma 2 9BChat
Qwen 2.5 VL 7BMultimodalLlama 3.1 8BChatQwen3 8BChatLearn to Run Local AI
Tutorials, demos and benchmarks from top channels
Ready to run?
Install Ollama and run any model with a single command. Click "ollama pull" on the cards above to copy the command.
Detect via Terminal
One command. Detects GPU, CPU and RAM, shows the top 3 models for you and opens the site with everything filled in.
npx can-i-run-aiZero dependencies
Uses only Node.js built-ins. Nothing to install besides npx.
Cross-platform
Windows, macOS and Linux. Detects hardware via native OS commands.
100% open source
Open source on GitHub. No data sent to servers.
Native apps — coming soon
Windows 10 and 11 — portable .exe
Apple Silicon & Intel — native app
Ubuntu, Fedora, Arch — AppImage
About CanIRunAI
CanIRunAI helps you find out which AI models can run on your computer. We detect your hardware automatically and calculate compatibility with dozens of popular open-source models.
Auto detection
GPU, RAM and CPU detected via WebGL and browser APIs.
Estimated performance
Tok/s and response time scaled by your GPU bandwidth.
100% local
No data leaves your browser. Everything runs client-side.
Compatible with Ollama, LM Studio and other local runtimes. Benchmark data based on real measurements with RTX 3060 as baseline.