Hosted Jupyter notebooks with optional free GPU and TPU runtimes
Google Colab runs Jupyter notebooks on Google hardware inside a browser tab, with no local install and no environment to maintain. Popular libraries for machine learning, data analysis and visualisation are preinstalled, datasets arrive from Google Drive or GitHub, and free sessions can include a GPU or TPU when capacity allows. Paid tiers buy longer runtimes, better accelerators and more memory for training. It is the quickest way to try a model or share a reproducible analysis, but not a production environment: sessions time out and compute is never guaranteed.
Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.
python experimentation, model fine-tuning, data analysis and teaching machine learning
If you're comparing similar products, check the alternatives below, or browse all tools in the AI Models & Platforms category.
No. Free accelerators are allocated by availability and can be unavailable at busy times, and sessions disconnect after inactivity or a maximum runtime. For repeated training runs, a paid plan or a rented cloud instance is far less frustrating.
Yes, through pip or apt inside the notebook, though installed packages disappear when the runtime resets, so put setup commands at the top of the notebook. Persistent storage is best kept in Drive.
Not really. It is an interactive environment without service guarantees, so anything that must stay online belongs on a proper server or a managed notebook service with long-running runtimes.
Developer console and API keys for the Claude model family
Node-based local interface for running image and video diffusion models
Browser playground for prompting and prototyping with Gemini models
Where most of the open machine learning world lives