Rent capacity, or provide it
The provider agent, escrowed rentals and on-chain settlement are already built into the platform. We're keeping the marketplace closed until inference demand from the API product justifies dedicated capacity — at which point your own GPUs can serve it.
NVIDIA H200
141 GB HBM3e
Frontier training & long-context inference
Pricing — · announced at launch
NVIDIA H100 SXM
80 GB HBM3
Production inference clusters
Pricing — · announced at launch
NVIDIA A100
80 GB HBM2e
Cost-efficient fine-tuning
Pricing — · announced at launch
NVIDIA RTX 5090
32 GB GDDR7
Single-node dev & rendering
Pricing — · announced at launch