Google Cloud Platform

Bringing Cloud ML Engine to more developers with online prediction features and reduced prices

Good news — we’ve adjusted our prices for, and added new features to Cloud Machine Learning Engine to help you do more with ML. Here’s what you can expect.

We’ve reduced our prices.

This will make a difference to the bottom line of over 10,000 paying customers using Cloud AI products.

  • Google Compute Engine GPU price reduction—We recently reduced the price of NVIDIA Tesla GPUs attached to on-demand Google Compute Engine virtual machines by up to 36 percent, allowing you to train your model faster and more cost effectively.
  • Cloud Machine Learning Engine price reduction—We introduced dramatic price reductions for the managed ML services (training, batch predict, online prediction)
  • Cloud ML Engine batch and online prediction (inference)—We applied the same price reductions and dropped the per-prediction fee to simplify billing. This lets you choose the best infrastructure for your production needs, without worrying as much about cost.

We’re introducing new features.

  • Online prediction is now generally available (GA): Cloud ML Engine online prediction is a service optimized to run your data through hosted models with as little latency as possible. You send small batches of data to the service and it returns your predictions in the response.
  • Online prediction alpha for scikit-learn + XGBoost: Please contact us if you are interested in testing as part of our early access program.
  • Python + TensorFlow updates:  Python 3 is now available for training as part of the Cloud ML Engine runtime version for TensorFlow 1.4.

Keep using what you already know and love.

For those of you building production quality machine learning into your applications, Cloud ML Engine provides a managed service so you can focus on your models, not infrastructure.

  • Focus on your TensorFlow model, not the infrastructure required to run the training, batch pipeline, or prediction servers
  • You only pay for what you use: you shouldn’t have to sacrifice choice for performance or price—on GCP you can choose your configuration and pay for what you use.
  • Easily enable large scale distributed training support: unlike other Cloud providers Cloud Machine Learning Engine doesn’t limit the dataset size for training
  • Easily scale web predictions for worldwide traffic of any reasonable volume, with state of the art spike handling
  • Easily run any number of batch predict jobs of any reasonable size for any model
  • TPUs are coming soon: with Cloud Machine Learning on TPUs you’ll be able to train and run machine learning models faster than before
We hope these updates to Cloud ML Engine unlock new opportunities and help you solve more interesting business, engineering, and scientific problems.

Check out our documentation for more information on Cloud ML Engine, take a look at the our pricing page, or try out our pricing calculator. Have questions or feedback? Let us know on our getting help page.

Getting started with Cloud ML Engine is easy: just start one up in the Google Cloud Console. If you don’t have a GCP account yet, sign up today and get $300 in credits.

One last thing—Cloud Machine Learning Engine limits resource allocation and use and enforces appropriate quotas on a per-project basis. If you need to increase quota, please contact us via the quota request form.