Google Collaboratory (Colab for short), the service of Google designed to allow anyone to write and run arbitrary Python code through a web browserIs introducing a pay-as-you-go plan. In its first price change since Google launched Colab premium plans in 2020, Colab will now give users the option to purchase additional computing time in Colab with or without a subscription.
Google says that the update will not affect Colab's free tier, which remains in its current form. The only material change is that users can purchase compute access in the form of "compute units," starting at $9.99 for 100 units or $49.99 for 500.
As Google Colab's product lead explains, Chris Perry, in a blog post:
Paid users now have the flexibility to use up their compute quota, measured in compute units, at a rate of their choosing. As compute units are depleted, a user may choose to purchase more on a pay-as-you-go basis at their discretion. Once a user has depleted their compute units, their Colab usage quota will revert back to our free tier caps.
Along with the pay-per-use launch, Google announced that paid Colab users can now choose between standard or "premium" GPUs in Colab; the latter are usually Nvidia V100 or A100 Tensor Core GPUs. (The standard GPUs in Colab are usually Nvidia T4 Tensor Core GPUs.) Nevertheless, The company notes that getting a specific GPU chip type assignment is not guaranteed and depends on several factors., including a user's availability and paid balance with Colab.
The sensitivity around price changes reflects how much Colab has grown since it emerged from an internal Google Research project in late 2017. The platform has become the de facto digital dashboard for demos within the AI research community; It is not uncommon for researchers who have written code to include links to Colab pages in or alongside the GitHub repositories that host the code.