📓

Google Colab

Google Colab (Colaboratory) is a free, browser-based Jupyter notebook environment made by Google, with no local setup and a free tier that includes GPU/TPU access. It ranks around #13 among collaboration tools with roughly 7% usage share, and it's the standard way students, researchers, and ML engineers share runnable Python notebooks — anyone with the link can open, run, and comment on the same notebook like a Google Doc. Teams and classrooms reach for it because it removes the "works on my machine" problem entirely.

Quick facts
Type: Cloud-hosted Jupyter notebook environment for Python
Made by: Google
Cost: Free tier (with usage limits); paid Colab Pro/Pro+ for more compute and longer GPU sessions
Best for: Sharing and collaboratively running Python/ML notebooks with zero local setup
Primary use case: Data science, machine learning experiments, and teaching Python with free GPU/TPU access

Jump to: Example Getting started Best for

Example

A Colab notebook is a sequence of cells. A common first cell installs a package that isn't preinstalled, then a code cell imports and uses it — here, checking for GPU access and running a small PyTorch tensor operation.

# Cell 1 — install a package not preinstalled in the Colab image
!pip install -q torch torchvision

# Cell 2 — confirm the free GPU runtime is attached, then run a quick tensor op
import torch

print("CUDA available:", torch.cuda.is_available())

x = torch.rand(3, 3).to("cuda" if torch.cuda.is_available() else "cpu")
y = x @ x.T
print(y)

Getting started

No install is required. Open Colab in any browser, create a new notebook, and switch on a free GPU/TPU runtime before running compute-heavy cells.

# 1. Go to colab.research.google.com and sign in with a Google account
# 2. File -> New notebook
# 3. Runtime -> Change runtime type -> select GPU (or TPU) -> Save
# 4. Share the notebook link the same way you'd share a Google Doc

Best for: Students, researchers, and small teams who want to write, run, and share Python/ML notebooks collaboratively without installing Python, CUDA, or Jupyter locally.