DocsBlogPricing

AI

The complete guide to Google Colab Compute Prices and Performance

Alec Fong

July 21, 20232 min read

Google Colab is a cloud-based notebook that provides access to CPU, GPU, and TPU resources. These resources can be used to train deep learning models, run data analysis, and perform other computationally intensive tasks. Let’s take a look at all the compute options that Google Colab has to offer.

Compute

As of July 2023

GPU TypeCPUSystem Memory (RAM)GPU Memory (VRAM)CostFree Tier
None (CPU Only)212GBNone$.008/hrYes
None (CPU Only)851GBNone$.019/hrNo
T4212GB16GB$.196/hrYes
T4851GB16GB$.224/hrNo
V100212GB16GB$.536/hrNo
V100851GB16GB$.564/hrNo
A1001283GB40GB$1.308/hrNo

Disk Storage

As of July 2023

AcceleratorFree TierPro Tier
None (CPU only)107 GB225 GB
GPU78 GB166 GB

Some datasets and workloads may require more disk storage. Take a look at how to manage your Google Colab state and storage for more information.

Network

As of July 2023. Testing shows the network can be highly variable.

NetworkMbit/s
Up300-800
Down600-2500

Conclusion

Overall, Google Colab provides a convenient and cost-effective way to access powerful computing resources for a wide range of tasks. While availability may vary, users can take advantage of usage statistics and monitoring tools to optimize their experience.

Learn how to best monitor these resources in Google Colab.

Previous
The No-BS Guide to Fine-Tuning an LLM