CanIRun.ai is a free, browser-based tool that instantly analyzes your computer's hardware to determine which AI models you can run locally. By leveraging browser APIs, it estimates your GPU, CPU, and RAM capabilities and matches them against a curated list of popular open-weight AI models. This eliminates the guesswork and technical barriers often associated with running large language models (LLMs) on personal machines.
Key features include automatic hardware detection, a comprehensive model database with over 30 models from leading organizations like Meta, OpenAI, Google, Alibaba, Mistral AI, and DeepSeek, and clear compatibility indicators. Models range from tiny 0.6B parameter variants for edge devices to massive 1T parameter MoE architectures. Each model entry displays its size, license, and a brief description, helping users choose based on their hardware limits and use case.
Benefits: Users save time by avoiding trial-and-error downloads of incompatible models. The tool provides realistic expectations about performance, as estimates are based on browser APIs and actual specs may vary. It empowers developers, researchers, and AI enthusiasts to experiment with local AI without expensive cloud subscriptions. Use cases include running chatbots, coding assistants, multimodal models, and reasoning engines on personal computers for privacy, offline access, or cost savings.
Technical details: The tool runs entirely in the browser, requiring no installation or data upload. It uses WebGL and navigator APIs to approximate GPU memory and compute capability, CPU cores and speed, and total system RAM. While not as precise as native system tools, it offers a quick and accessible starting point. The model database is regularly updated to include the latest releases. Note that running large models may still require significant resources, and the tool's recommendations are estimatesโactual performance depends on your specific hardware configuration and system load.
developers, researchers, AI enthusiasts, data scientists, machine learning engineers, hobbyists