Product & Pricing
Bring your concept, existing data and AI Skills.
Experience the Instant Super Computer.
Ops.
Manage your expenses effectively with our cost control feature. Set maximum budgets per day or month for your organization, each project, and per engineer. Enjoy the peace of mind with real-time consumption tracking, accurate to about one minute.
Included at no additional charge.
AI Datasets.
Storage. Transfer.
2x Global Replicated at $0.05 per GB per month.
Dataset Loading:Reserved Workstation: $0,
Speed:5-20GB (40-160Gbps) per second.
Development Cluster: $0
Speed:5-20GB (40-160Gbps) per second.
Training Cluster: $0.02 per GB
Speed: 1-20GB (8-160Gbps) per second.
Checkpoint Transit $0.10 per GB.
Dataset Loading:Reserved Workstation: $0,
Speed:5-20GB (40-160Gbps) per second.
Development Cluster: $0
Speed:5-20GB (40-160Gbps) per second.
Training Cluster: $0.02 per GB
Speed: 1-20GB (8-160Gbps) per second.
Checkpoint Transit $0.10 per GB.
Reserved GPU Workstation.
Your new terminal.
Free, 20GB(160Gbps) per second access to your data.
Free, 20GB(160Gbps) per second access to your data.
Coming soon...
Development Cluster.
Launch in 10s. Validate your distributed training with rapid feedback.
No dataset / checkpointing transit fees.
Shared cluster, Jobs typically start in 10 seconds.
No dataset / checkpointing transit fees.
Shared cluster, Jobs typically start in 10 seconds.
$10 per 90 seconds (72x24GB Ampere GPU Cluster).
Training Cluster.
Launch in 10mins.
Dedicated cluster to run your job until completion.
On Demand pricing - only pay when you use it.
Running on our multicloud backend and global data network to maximize availability.
Run as many as you want in parallel.
Real time logging with your standard tooling - WandB, TensorBoard and more.
Dedicated cluster to run your job until completion.
On Demand pricing - only pay when you use it.
Running on our multicloud backend and global data network to maximize availability.
Run as many as you want in parallel.
Real time logging with your standard tooling - WandB, TensorBoard and more.
$25 start fee + $4 per minute (72x 24GB Ampere GPU cluster)
Dev Libraries.
Open source libraries for: Robust state saving, well beyond simple checkpointing and working examples for common models.
Included at no additional charge.
Leave with trained models and data science outputs.
Without us
With Strong Compute
MLOps, cluster setup
Months
Less than 1 hour
Time AI Developers waiting for cluster to be ready
Months
Immediate (0)
Queuing, multi-cloud
Months to years
Included
Dataset Storage
Varies
$0.05 per GB per month (billed by the minute)
Data Transfer
Varies
$0.012 for data science, $0.10 per GB for checkpoints
Single GPU development
Varies
Coming soon
Develop Multinode Training
Months, thousands of dollars
$10 for 72x24GB GPUs per 90 seconds with rapid launch and inclusive pricing
Our 'Cycling' option revolutionizes multinode training with a launch time of just 10 seconds. It costs only $10 per 90-second cycle, leveraging a powerful 72x 24GB Ampere GPU cluster. Enjoy the added benefit of no data transit costs, making it a cost-effective solution for rapid development needs.
Multinode Training
Months of setup, queuing, variable costs
Starts at $25 launch and $4 per minute for 72x24GB GPUs + dataset transit & checkpoint transit
The 'Burst' mode is designed for efficient and fast multinode training, with a cluster setup time of less than 10 minutes. It offers transparent pricing at $25 per job launch and $4/min for a 72 GPU cluster. Dataset transit is economically priced at $0.012/GB, and checkpoint services are available at $0.095/GB, ensuring a clear understanding of costs involved.
Cost Controls
Typically opaque, costs known only after it's too late
Set fixed limits with real-time tracking
Manage your expenses effectively with our cost control feature. Set maximum budgets per day or month for your organization, each project, and per engineer. Enjoy the peace of mind with real-time consumption tracking, accurate to about one minute.
