Unified control plane for managing computational workloads across any infrastructure
Stage
Not given
Established
Founded
Not given
2015
Based in
US
US
Pricing model
Paid
Paid
Pricing
CPU/RAM share: $3 per hour per whole CPU node; GPU/VRAM share: $6 per hour per full GPU card. Rates are set by individual hosts and metered by the second. App publisher fees apply to paid apps. Payments via USDC on Base network or card via Stripe.
Per-user licensing recommended for up to 100 end users, program licensing above that. Program license is a flat annual fee set by deployment location (shared platform tenant, dedicated cloud, or dedicated on-premises). Academic institutions receive 50% off program and per-user licenses. Five-year te
Key features
WebAssembly portability across backends
Hardware attestation verification in browser
CPU isolation with Enclave Shield
GPU acceleration with masked offload
Per-second billing
AMD SEV-SNP confidential processor support
Encrypted volumes for apps
REST API and CLI control
On-chain deployment records
Decentralized host network
Unified control plane for on-premises and cloud computing resources
Support for Slurm and Kubernetes schedulers
Flat annual licensing with unlimited users and resources
Enterprise features including SSO and user impersonation
Budget and chargeback management
Integration with cloud providers and on-premises infrastructure
Comprehensive API and CLI for programmatic access
What makes it different
Only platform where hosts prove protection before receiving data
Portable apps work identically across all supported backends
Non-custodial: deployments on public blockchain, no company control
Pay only for actual compute used with per-second metering
Designed to outlive any operator through decentralized architecture
No per-user, per-GPU, or consumption charges on program licenses
Included mission engineering team sizes based on usage, not user counts
50% academic pricing discount
Use existing cloud accounts and GPU contracts at no additional charge
Enables workload shifting across clusters and clouds without license changes