TurboML runs on fully managed infrastructure that scales with you.
Start building today with product and support plans tailored to your needs.
For growing businesses
Billed on the basis of usage
Ideal for: Small to medium-sized businesses scaling machine learning operations.
For large enterprises
Contact sales for more info
Ideal for: Large enterprises with complex ML workloads and hybrid infrastructure requirements.
| Features | Pro | Enterprise |
|---|---|---|
| Access to CPU & GPU Instances | Supported | Supported |
| Triggers & Automations | Supported | Supported |
| Experiment Tracking | Supported | Supported |
| Cost & Usage Tracking | Supported | Supported |
| Operational & Model Deployment Features | Supported | Supported |
| Host, Serve & Monitor Custom Models | Supported | Supported |
| Custom Metrics | Supported | Supported |
| SLA for Critical Cases (<3 hrs) | Supported | Supported |
| Priority Technical Support | Supported | Supported |
| Audit Trails | Not supported | Supported |
| Custom SLA | Not supported | Supported |
| RBAC & SSO Support | Not supported | Supported |
| Advanced Security & Compliance | Not supported | Supported |
| Self-Hosted Hybrid Deployments | Not supported | Supported |
| Customizable TBU Packages | Not supported | Supported |
| 200+ Data Sources & Custom Connectors | Not supported | Supported |
| Dedicated Account Manager | Not supported | Supported |
| Advanced Security & Compliance (SOC 2, GDPR, HIPAA) | Supported | Supported |
| Instance | CPU | Memory | TBU |
| tiny | 1 | 2 GB | 25 |
| small | 2 | 8 GB | 50 |
| medium | 4 | 16 GB | 100 |
| large | 8 | 32 GB | 200 |
| xlarge | 16 | 64 GB | 400 |
| 2xlarge | 32 | 128 GB | 800 |
| 4xlarge | 64 | 256 GB | 1600 |
| 8xlarge | 128 | 512 GB | 3200 |
| 12xlarge | 192 | 768 GB | 4800 |
| Instance | GPU Type | GPUs | CPU | Memory | TBU |
| gpu1.small | NVIDIA A10G | 1 | 4 | 16 GB | 150 |
| gpu1.medium | NVIDIA A10G | 1 | 8 | 32 GB | 180 |
| gpu1.large | NVIDIA A10G | 1 | 16 | 64 GB | 240 |
| gpu1.xlarge | NVIDIA A10G | 1 | 32 | 128 GB | 375 |
| gpu1.2xlarge | NVIDIA A10G | 4 | 48 | 192 GB | 700 |
| gpu2.small | NVIDIA T4 | 1 | 4 | 16 GB | 75 |
| gpu2.medium | NVIDIA T4 | 1 | 8 | 32 GB | 100 |
| gpu2.large | NVIDIA T4 | 1 | 16 | 64 GB | 180 |
| gpu3.small | NVIDIA L4 | 1 | 4 | 16 GB | 120 |
| gpu3.medium | NVIDIA L4 | 1 | 8 | 32 GB | 150 |
| gpu3.large | NVIDIA L4 | 1 | 16 | 64 GB | 190 |
| gpu3.xlarge | NVIDIA L4 | 1 | 32 | 128 GB | 300 |
| gpu4.large | NVIDIA A100 | 1 | 4 | 61 GB | 150 |
| gpu4.8xlarge | NVIDIA A100 | 8 | 32 | 488 GB | 1000 |
| gpu5.48xlarge | NVIDIA H100 | 8 | 192 | 2048 GB | 15000 |
TurboML Billing Units (TBU) Pricing Model
TurboML provides a straightforward and flexible billing system based on TurboML Billing Units (TBU), allowing you to pay only for what you use. This predictable pricing model adapts to the needs of organizations of all sizes, ensuring cost efficiency.
What is a TurboML Billing Unit (TBU)?
A TurboML Billing Unit (TBU) is the core metric for usage-based billing on the TurboML platform. It represents the normalized cost of operations such as data processing, storage, and retrieval. Your monthly costs are calculated based on the number of TBUs consumed across various activities like retrieval, and storage.
1 TBU = $0.01
Your monthly charges will depend on the number of TBUs consumed across various operations such as data ingestion, data retrieval, and data storage.
How TBUs are Consumed
1. Data Retrieval (Reads):
Accessing or querying data from TurboML incurs TBUs based on the size of the data retrieved.
Example: Retrieving 1 GB of data may cost 500 TBUs.
2. Data Storage:
Storing data on the TurboML platform is charged based on the volume stored and the storage duration.
Example: Storing 1 GB of data per hour may cost 0.05 TBUs.
3. Instance Types:
Running workloads on TurboML involves charges for compute resources based on instance types. Larger instances consume more TBUs per hour.
Billing Examples with Instance Types
Scenario 1: Small-Scale Use Case (General-Purpose Instances)
A small organization deploys eight "Large" instance types at a rate of 200 TBUs/hour.
• Instance Usage: 8 instances x 200 TBUs/hour x 24 hours x 30 days
• Estimated TBUs Used: 1,152,000 TBUs
Scenario 2: Enterprise Use Case (General-Purpose Instances)
An enterprise deploys sixteen "2xLarge" instance types at a rate of 800 TBUs/hour.
• Instance Usage: 16 instances x 800 TBUs/hour x 24 hours x 30 days
• Estimated TBUs Used: 9,216,000 TBUs
Fully managed by TurboML on secure servers, giving you a seamless and worry-free trial experience.
Contact Sales