Energy cost per GPU-hour by US power region + regional basis (no params)

Answeringour last check, 2026-09-24

1 of 1checks answered this week

833 msmedian answer time

$0.1listed price per call

$0.1price it asked us

Paid test badge: not yet. The checks above are free: we call the tool without paying and read the payment request it sends back. The Verified badge needs paid calls whose answers match the promised output, and nobody can buy a badge.

Endpoint

GET https://computeindex.lonestaroracle.xyz/index

Category	Everything else
Provider host	computeindex.lonestaroracle.xyz
Networks	eip155:8453, solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp
Payment schemes	exact
Self-reported calls, 30 days	0 from 0 payers (the provider's figure, not ours)

Our checks, last 30 days

Day	Result	HTTP	Asked	Time
2026-09-24	valid payment request	402	$0.1	833 ms

Example input (from the provider)

{
  "method": "GET",
  "type": "http"
}

Promised output schema (from the provider)

{
  "category": "compute",
  "description": "The regional cost of powering AI compute, made agent-callable. One x402 call returns the energy cost to run a GPU for one hour (H100, H200, B200, A100) in each US wholesale power region (CAISO SP15/NP15; ERCOT Houston, North, South, West, hub average), each region's basis versus the average, the cheapest and priciest region, the spread translated to a 1,000-GPU cluster per month, live GPU spot rental prices (Vast.ai, RunPod, AWS spot, Lambda reference) and grid stress (EIA-930). Assumptions are published in every response (GPU board kW x PUE x real-time $/MWh). Compute is becoming an economic resource with regional pricing; this is the price-discovery input a data-center siting, workload-scheduling, compute-procurement or compute-finance agent needs. $0.10 in USDC on Base or Solana via x402. Built on LSO ComputePulse + SparkPulse + GridPulse.",
  "name": "ComputeIndex",
  "properties": {
    "input": {
      "properties": {
        "method": {
          "enum": [
            "GET",
            "HEAD",
            "DELETE"
          ],
          "type": "string"
        },
        "queryParams": {
          "properties": {},
          "type": "object"
        },
        "type": {
          "const": "http",
          "type": "string"
        }
      },
      "required": [
        "type",
        "method"
      ],
      "type": "object"
    },
    "output": {
      "properties": {
        "example": {
          "assumptions": {
            "gpu_board_kw": {
              "B200": 1,
              "H100": 0.7
            },
            "pue": 1.3
          },
          "cheapest_region": "ERCOT-West",
          "gpu_spot_avg_per_hr": {
            "A100": 2.12,
            "H100": 2.46,
            "H200": 2.05
          },
          "priciest_region": "CAISO-NP15",
          "regions": [
            {
              "basis_vs_avg_per_H100_hr": -0.00767,
              "energy_cost_per_gpu_hr": {
                "B200": 0.01872,
                "H100": 0.0131
              },
              "energy_share_of_spot_pct": {
                "H100": 0.53
              },
              "grid_stress": "HIGH",
              "market": "ERCOT",
              "region": "ERCOT-West",
              "rt_mwh": 14.4
            }
          ],
          "signal": "WIDE_BASIS",
          "spread_per_1000_gpu_cluster_month_usd": 13702,
          "spread_per_H100_hr": 0.01877,
          "summary": "Cheapest power for compute right now: ERCOT-West at $14.4/MWh ($0.0131/H100-hr in energy); priciest: CAISO-NP15 at $35.02/MWh. Spread = $0.01877/H100-hr, about $13,702/month for a 1,000-GPU cluster."
        },
        "type": {
          "type": "string"
        }
      },
      "required": [
        "type"
      ],
      "type": "object"
    }
  },
  "required": [
    "input"
  ],
  "tags": [
    "compute",
    "gpu",
    "electricity",
    "energy",
    "ai-infrastructure"
  ],
  "type": "object",
  "version": "1.0.0"
}

This page as JSON