We are excited to announce new pricing for training custom tabular models in Amazon SageMaker Canvas, a no-code tool that enables customers to easily create highly accurate ML models without writing code. SageMaker Canvas supports numeric prediction (regression), 2 category prediction (binary classification), 3+ category prediction (multi-class classification), and time-series forecasting for tabular models. Previously, model training charges were based on the number of cells in the dataset used to train the model. Now, the charges are based on SageMaker training and processing hours used to train the model.ENnbsp;

The new pricing is cost-effective, enables quick iteration, and provides consistent pricing to train custom models across different modalities in SageMaker Canvas. Previously, model training in Canvas incurred a minimum of $30, and the cost increased with higher cell count. Now, the training charges for tabular models are based solely on SageMaker instance hours used, thereby significantly lowering the overall training cost. For example: A quick build model with 16 MB of data with about 3 million cells can now be as low as under $2, and a standard build model, which runs a comprehensive AutoML experiment may use 2 ml.c5.18xlarge instance hours and still cost you only $7.30.ENnbsp;

The updated pricing is now available in all AWS regions where SageMaker Canvas is supported. You will see SageMaker training and processing charges for various instance types, including ml.m5.12xlarge, ml.c5.18xlarge, and ml.m5.4xlarge, that Canvas automatically selects based on performance and availability. The pricing for these instances can be found at Amazon SageMaker pricing. To learn more, refer to SageMaker Canvas Pricing.