macrostack
Tool profile · Cloud GPU & AI Compute

Kaggle Notebooks

Named free GPUs and a free TPU, in sessions of up to 12 hours.

30
sovereignty

Kaggle's notebooks give free GPUs (a single P100 or two T4s) and a free TPU v3-8, in sessions of 12 hours on CPU and GPU and 9 on TPU, with 20GB of saved output and a 20-minute idle limit. Free GPU time is limited, and busy periods put you in a queue. Colab Pro and Pro+ users get 15 and 30 extra GPU hours a week there, a limited-time promotion. Kaggle's Jupyter Server can also be reached from Colab or VS Code.

SOURCE-AVAILABLEProprietary (hosted service)
LicenseProprietary (hosted service)
PricingFree.
Open sourceNo
Self-hostableNo
Local-first dataNo

What it does well

  • +Named free GPUs: a P100, or two T4s
  • +A free TPU v3-8
  • +Sessions of up to 12 hours, as long as Colab's free maximum

Where it falls short

  • −Free GPU time is limited, with a queue when busy
  • −4 CPUs and about 30GB of RAM per session
  • −Hosted and proprietary: no way to run it yourself

Kaggle Notebooks as an alternative to

Where Kaggle Notebooks shows up in our comparisons, and how it ranked.

Kaggle Notebooks head-to-head

Straight comparisons against the tools people weigh it against.

The Macrostack brief

New swaps, worth your inbox.

A short, occasional email when we add a high-intent alternative or ship a new head-to-head. No spam, no selling your address — unsubscribe in one click.