What is canirunai?

CanIRunAI is a small local web app that answers a simple question: which AI models can your own computer actually run. It scans your CPU, GPU, VRAM, and system RAM when the server starts, then compares that against a built in list of over sixty models, covering chat and coding language models, image generators, and video generators. Each model gets a plain verdict of can run, tight fit, or too heavy, based on your real hardware rather than generic minimum requirements. The README describes it as similar in spirit to the older "Can You Run It" game checking sites, but for AI models instead of video games. It pays specific attention to Apple Silicon Macs, calculating how much of their shared unified memory is effectively available as VRAM, and it separately highlights models that can run on CPU alone with no graphics card at all. If you have Ollama installed, the app can also pull models directly from its interface with a live progress bar, though Ollama is optional and only needed for that pulling feature, not for the compatibility checking itself. Running it locally is a standard Node.js setup: clone the repository, install dependencies, run the dev command, then open the app in your browser. There is no build step or bundler involved. The backend is a small Express server that does the hardware detection and proxies requests to Ollama, while the frontend is a single plain HTML page with no JavaScript framework. The project welcomes contributions such as adding more models to its list or improving GPU detection on Linux, and it is released under the MIT license.