Our technology handles the entire training process for your models and only charges you for the resources you’ve used to millisecond precision. No more worrying about idle instances or hidden fees.
Our pricing structure is aimed to be easy to understand and completely transparent so you always know where you stand and your engineers can just focus on Machine Learning.
When using GPUs on our platform, up to 8 CPUs will be available to your training job free of charge, please get in touch if you have more specific requirements.
We currently offer a maximum of 8 GPUs per training job.
|CPU||$0.06 per Hour||Intel Xeon Processors|
|GPU||$3.99 per Hour||NVIDIA Tesla V100|
|Long Term Storage||$0.029 per GB-month||Charged per data set|
|Ephemeral Storage (SSD)||$0.00021 per GB-hour||Charged per model trained, ~$0.154 per GB-month|
Long Term Storage: Your original upload and source of truth for a data set.
Ephemeral Storage: Each trained model is paired with a short lived copy of your original data set, local to the training operation, ensuring your models are trained as fast as possible.*
* Your data will always remain in the region you uploaded it to.
** We can begin serving any region quickly. Please get in touch if you require a specific region.