Runs on our shared community cluster
Execute, monitor, and track ML experiments with simple, yet efficient tools
Get $10 of GPU credit when you sign up for a new account
Installs in seconds in your private AWS, GCP, or Azure account
Includes collaboration features for coordinating work across ML/DS teams
Supports execution on private hardware (e.g. nVidia DGX workstations)
Includes Kubernetes-based autoscaling Model Deployment and Monitoring
Custom deployed in multi-cloud, hybrid-cloud, or fully on-prem configurations
Enterprise-grade security, SSO, and user/data access control
Solutions team to help with data management, model development, and deployment
Execute training runs utilizing git-integrated cloud compute in your local CLI
Manage datasets, model artifacts, and other resources, and integrate your own cloud accounts within our Spell filesystem
Define and schedule end-to-end processes from training to deployment for automation and reproducibility
Create hosted Jupyer environments in the cloud, accessible from anywhere
Leverage grid, random, and bayesian hyperparameter searches and visualize your results
Distribute DL training loads with our native Horovod integration
Compare and visualize training, hardware, and other custom metrics and metadata within our web console
Utilize our CLI tools and Python API to customize logging and tracking
Invite team members and organize team projects within your private organization account
Create models from best performing training runs and record their artifacts to prepare for deployment
Serve single or multiple trained models in real-time or batch, backed by a managed Kubernetes service for maximum scalability and robustness
Track end-to-end lineage and metadata from dataset to deployed server, and reproduce any model backed by Spell runs as needed
Make use of existing Google accounts for authentication and other necessary security features for your organization (e.g. multi-factor authentication)
Control data access and user privileges for your organization from our web dashboard
Configure Spell around an existing cloud provider account, backed entirely by your own VPC and compute instances
Integrate your existing enterprise authentication system with Spell's services (e.g. LDAP, Active Directory)
Integrate your existing private storage volumes (e.g. NFS, HDFS) along with customized data access controls
Use Spell's public cloud GPU clusters for training models
Deploy Spell within your own public-cloud backed account, managed within a VPC
Host Spell fully on-prem within your own bare-metal, private cloud, or other infrastructural setup
Join our public Slack channel for questions, comments, and feedback, directly to our Engineering teams
Use a private Slack channel for your organization for dedicated issue support
Obtain white glove service backed by our enterprise level SLA
Hire our experienced ML solutions engineers for various deep learning advisory services
We use Stripe to process payments. We accept all major credit and debit cards including Visa, MasterCard, American Express and Discover. We do not currently accept prepaid cards.
Check if you’re using an accepted credit or debit card (Visa, MasterCard, American Express, and Discover). We use Stripe to process payments. If you’re still unable to add your credit card, contact us at support@spell.ml.
Billing occurs monthly and starts one month after your sign up date. You can check your billing period on your settings page or by using the spell status command.
Yes. We bill by time used, so you will still be billed if you kill your run or if your run fails.