Broker
Usage markupERA routes workloads to the cheapest viable provider and adds a transparent margin.
Best for teams that want one bill and zero provider setup.ERA Cloud compares GPU, VPS/CPU, AI inference and cloud capacity across 40+ providers, then routes workloads by price, region, availability, and policy. The customer sees one API, one dashboard, and one operating model.
ERA routes workloads to the cheapest viable provider and adds a transparent margin.
Best for teams that want one bill and zero provider setup.Bring provider keys, keep direct provider billing, and pay ERA for routing and control-plane automation.
Best for teams already using AWS, GCP, Yandex, RunPod, Vast.ai, or similar.Use your preferred providers first, then burst to cheaper external pools when capacity or price changes.
Best for production teams with compliance, geography, or partner constraints.Public self-service checkout is intentionally disabled while ERA Pay, provider clearing, refunds, and payout workflows are finalized. Beta customers receive written terms, API keys, usage limits, and provider routing policies during onboarding.
These are planning ranges for sales and onboarding. Live routing should use the benchmark endpoint and provider adapters after real credentials are configured.
| GPU | ERA low target | Typical cloud band | Best use |
|---|---|---|---|
| H100 | $1.30/h | $2.60-4.10/h | frontier inference, fine-tuning, batch jobs |
| A100 | $0.55/h | $1.40-3.20/h | training, embeddings, stable diffusion |
| L40S | $0.42/h | $0.90-1.80/h | LLM inference, video, rendering |
| RTX 4090 | $0.22/h | $0.45-1.10/h | prototyping, image generation, experiments |
| T4 | $0.09/h | $0.20-0.55/h | small inference, dev environments |