Public API

Versioned · no auth · JSON · generatedAt timestamp

Aggregation Marketplace exposes a versioned, read-only JSON API. No API key is required. Every response carries a version and a generatedAt field with the real ISO timestamp of the response.

GET /api/v1/gpus

Every canonical GPU model currently tracked across cloud providers, with provider/instance counts, per-GPU price range, and hardware specs.

FieldTypeDescription
canonicalstringCanonical display name (e.g. "H100 80GB")
slugstringCanonical slug used on model pages (/gpu/<slug>)
providerCountnumberDistinct providers offering the model
instanceCountnumberAvailable configurations/instances
priceMinCentsnumberCheapest per-GPU per-hour price, USD cents
priceMaxCentsnumberMost expensive per-GPU per-hour price, USD cents
spec.vramnumber|nullVRAM in GB
spec.memoryTypestring|nullMemory type (HBM3, GDDR6X, …)
spec.bandwidthnumber|nullMemory bandwidth in GB/s
spec.interconnectstring|nullInterconnect — not tracked yet (null)
curl https://aggregation.marketplace/api/v1/gpus

GET /api/v1/providers

Cloud providers with live GPU offerings: aggregate price stats per provider, regions, VRAM range, and egress pricing when published by the provider.

FieldTypeDescription
slugstringProvider slug
namestringProvider display name
websitestring|nullProvider website
countrystring|nullProvider country
gpuTypesnumberDistinct GPU models offered
configurationsnumberAvailable configurations/instances
regionsnumberDistinct regions
priceMinCentsnumberCheapest per-GPU per-hour price, USD cents
priceMedianCentsnumberMedian per-GPU per-hour price, USD cents
priceMaxCentsnumberMost expensive per-GPU per-hour price, USD cents
vramMin / vramMaxnumber|nullVRAM range across offerings
egressCostPerGbnumber|nullEgress cost per GB in USD (null when unpublished)
egressFreeGbnumber|nullFree egress allowance in GB (null when unpublished)
egressSourceUrlstring|nullSource URL for the egress tariff
lastCheckedAtstring|nullLast successful check ISO timestamp
curl https://aggregation.marketplace/api/v1/providers

Example response

{
  "version": "v1",
  "generatedAt": "2026-09-03T12:00:00.000Z",
  "count": 3,
  "gpus": [
    {
      "canonical": "H100 80GB",
      "slug": "h100-80gb",
      "providerCount": 7,
      "instanceCount": 105,
      "priceMinCents": 79,
      "priceMaxCents": 6016,
      "spec": {
        "vram": 80,
        "memoryType": "HBM3",
        "bandwidth": 3350,
        "interconnect": null
      }
    }
  ]
}

Prices are normalized to USD cents per GPU per hour so configurations with multiple GPUs remain comparable. The legacy unversioned endpoints (/api/listings, /api/gpu/*) stay available; new integrations should use the v1 surface.