Generally available
GPUs
On-demand accelerators, by the second.
Open sourceOn-chain settlement
- API
- /v1/gpus
- Google Cloud
- Compute Engine GPUs
H100, A100, and consumer-class GPUs on demand — attach to a machine, a container, or a training job. Per-second billing, no reservations, no idle spend.
H100 / A100 / L40S
Per-second metering
Attach to any workload
No reservation required
The AI cloud — one API for every capability. Open models. On-chain.