跳至正文
MZGrid

此页面仅提供英文版本。 以英文版本为准。

Public endpoints

Unauthenticated endpoints for health, the GPU and template catalogue, marketplace offers and live network statistics.

These endpoints require no authentication and power the marketplace and network dashboard. They are rate-limited per IP address.

GET /v1/health

Service health, the chain the coordinator is connected to and the latest indexed block.

JSON
{ "status": "ok", "chainId": 8453, "block": 22917384, "version": "1.6.0" }

GET /v1/catalog/gpus

The GPU catalogue with live supply and price statistics per SKU.

JSON
[
  {
    "sku": "h100-sxm",
    "name": "NVIDIA H100 SXM",
    "vramGb": 80,
    "arch": "Hopper",
    "tflopsFp16": 989,
    "referencePrice": "750.000000",
    "availableGpus": 61,
    "totalGpus": 336,
    "lowestPrice": "650.000000",
    "medianPrice": "755.000000"
  },
  {
    "sku": "rtx4090",
    "name": "GeForce RTX 4090",
    "vramGb": 24,
    "arch": "Ada",
    "tflopsFp16": 165,
    "referencePrice": "150.000000",
    "availableGpus": 402,
    "totalGpus": 1456,
    "lowestPrice": "130.000000",
    "medianPrice": "152.000000"
  }
]
FieldTypeDescription
skustringStable GPU identifier.
namestringMarketing name.
vramGbnumberVRAM in GB.
archstringArchitecture: Blackwell, Hopper, Ada or Ampere.
tflopsFp16numberFP16 dense TFLOPS benchmark reference.
referencePricestringCatalogue reference price, MZG per GPU-hour.
availableGpus / totalGpusnumberGPUs of this SKU currently free / listed.
lowestPrice / medianPricestringAcross live offers, MZG per GPU-hour.

GET /v1/catalog/templates

JSON
[
  {
    "id": "llama-3.1-8b",
    "name": "Llama 3.1 8B Instruct (vLLM)",
    "kind": "inference endpoint",
    "description": "Llama 3.1 8B Instruct served by vLLM with an OpenAI-compatible API.",
    "minVramGb": 24,
    "image": "vllm/vllm-openai:v0.6.6",
    "ports": [8000],
    "recommendedSkus": ["rtx4090", "rtx3090", "rtx5090", "l40s"]
  },
  {
    "id": "pytorch-jupyter",
    "name": "PyTorch 2.5 + CUDA 12.4 JupyterLab",
    "kind": "interactive",
    "description": "JupyterLab with PyTorch 2.5 and CUDA 12.4.",
    "minVramGb": 8,
    "image": "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime",
    "ports": [8888],
    "recommendedSkus": ["rtx4090", "a100-80"]
  }
]

kind is one of inference endpoint, image generation, interactive, custom or batch. See Templates.

GET /v1/market/offers

Live marketplace offers.

QueryTypeDescription
skustringFilter by SKU, e.g. h100-sxm.
regionstringFilter by region code, e.g. eu-central.
minVramnumberMinimum VRAM in GB.
minReliabilitynumberMinimum provider reliability, 0–100.
sortstringprice (ascending), reliability or availability (descending).
limitnumberPage size.
cursorstringnextCursor from the previous page.
Shell
curl "https://api.mzgrid.io/v1/market/offers?sku=h100-sxm&minReliability=95&sort=price&limit=1"
JSON
{
  "items": [
    {
      "id": "off_4tq8n1zv6c",
      "provider": {
        "address": "0x8b3f…2e41",
        "name": "kestrel-compute",
        "tier": "datacenter",
        "reliability": 98,
        "uptime30d": 99.7,
        "jobsCompleted": 4182,
        "memberSince": "2026-09-11T14:03:00Z"
      },
      "sku": "h100-sxm",
      "gpuName": "NVIDIA H100 SXM",
      "gpuCount": 8,
      "vramGb": 80,
      "cpuCores": 112,
      "ramGb": 1024,
      "diskGb": 7680,
      "bandwidthMbps": 10000,
      "region": { "code": "eu-central", "country": "DE", "city": "Frankfurt" },
      "pricePerHour": "745.000000",
      "available": 3,
      "inUse": 5,
      "verified": true,
      "lastBenchmarkAt": "2026-10-29T03:20:11Z",
      "benchmarkScore": 962.4
    }
  ],
  "nextCursor": "eyJwIjoiNzQ1IiwiaWQiOiJvZmZfNHRxOG4xenY2YyJ9"
}

Offer object

FieldTypeDescription
idstringOffer id; pass as offerId to POST /v1/jobs.
provider.addressstringProvider wallet address.
provider.namestringDisplay name.
provider.tierstringdatacenter, pro or community.
provider.reliabilitynumberReputation score 0–100.
provider.uptime30dnumberUptime over 30 days, percent.
provider.jobsCompletednumberLifetime completed jobs.
provider.memberSincestringRegistration time.
sku, gpuName, vramGbGPU model.
gpuCountnumberGPUs in the machine: 1, 2, 4 or 8.
cpuCores, ramGb, diskGb, bandwidthMbpsnumberMachine specification.
regionobjectcode, country (ISO 3166-1 alpha-2) and city.
pricePerHourstringMZG per GPU-hour.
available / inUsenumberGPUs free / currently running jobs.
verifiedbooleanBenchmark within 70–115% of the SKU reference.
lastBenchmarkAtstringTime of the most recent benchmark.
benchmarkScorenumberMost recent measured FP16 TFLOPS per GPU.

GET /v1/market/offers/:id

A single offer with its price history and recent jobs.

JSON
{
  "id": "off_4tq8n1zv6c",
  "provider": {
    "address": "0x8b3f…2e41",
    "name": "kestrel-compute",
    "tier": "datacenter",
    "reliability": 98,
    "uptime30d": 99.7,
    "jobsCompleted": 4182,
    "memberSince": "2026-09-11T14:03:00Z"
  },
  "sku": "h100-sxm",
  "gpuName": "NVIDIA H100 SXM",
  "gpuCount": 8,
  "vramGb": 80,
  "cpuCores": 112,
  "ramGb": 1024,
  "diskGb": 7680,
  "bandwidthMbps": 10000,
  "region": { "code": "eu-central", "country": "DE", "city": "Frankfurt" },
  "pricePerHour": "745.000000",
  "available": 3,
  "inUse": 5,
  "verified": true,
  "lastBenchmarkAt": "2026-10-29T03:20:11Z",
  "benchmarkScore": 962.4,
  "priceHistory": [
    { "t": "2026-10-21T00:00:00Z", "price": "760.000000" },
    { "t": "2026-10-28T00:00:00Z", "price": "750.000000" },
    { "t": "2026-11-04T00:00:00Z", "price": "745.000000" }
  ],
  "recentJobs": [
    { "template": "llama-3.3-70b", "durationS": 14400, "status": "completed", "endedAt": "2026-11-04T07:41:09Z" },
    { "template": "batch", "durationS": 5312, "status": "completed", "endedAt": "2026-11-04T05:02:47Z" }
  ]
}

GET /v1/network/stats

Network-wide totals, refreshed continuously.

JSON
{
  "providers": 1284,
  "gpusOnline": 4917,
  "gpusTotal": 5612,
  "countries": 31,
  "utilization": 70.6,
  "jobsRunning": 3468,
  "jobsCompleted24h": 3127,
  "jobsCompletedTotal": 143071,
  "gpuHoursTotal": 1968240,
  "gpuHours24h": 58240,
  "mzgSettledTotal": "343245924.210000",
  "mzgSettled24h": "9019936.250000",
  "mzgLocked": "63466925.020000",
  "activeCustomers30d": 3911,
  "avgReliability": 93.2,
  "launchedAt": "2026-08-06T09:00:00.000Z",
  "networkAgeDays": 90,
  "updatedAt": "2026-11-04T09:15:00Z"
}

utilization and avgReliability are percentages; mzg* fields are MZG amounts. launchedAt is the network's mainnet launch and networkAgeDays the whole days since then.

GET /v1/network/timeseries

QueryValues
metricgpuHours, jobs, providers, gpusOnline, mzgSettled, utilization
range7d, 30d, 90d, 180d

No points are returned from before launchedAt: for a range longer than the network's age, only the available buckets are returned (the first one may start at the launch).

JSON
{
  "metric": "gpuHours",
  "range": "7d",
  "points": [
    { "t": "2026-10-29T00:00:00Z", "v": 55310 },
    { "t": "2026-10-30T00:00:00Z", "v": 56102 },
    { "t": "2026-10-31T00:00:00Z", "v": 56874 },
    { "t": "2026-11-01T00:00:00Z", "v": 55988 },
    { "t": "2026-11-02T00:00:00Z", "v": 57140 },
    { "t": "2026-11-03T00:00:00Z", "v": 57902 },
    { "t": "2026-11-04T00:00:00Z", "v": 58240 }
  ]
}

GET /v1/network/regions

JSON
[
  { "code": "us-east", "name": "US East", "country": "US", "providers": 168, "gpus": 812, "utilization": 72.4, "lat": 39.04, "lon": -77.49 },
  { "code": "eu-central", "name": "EU Central", "country": "DE", "providers": 121, "gpus": 566, "utilization": 69.8, "lat": 50.11, "lon": 8.68 },
  { "code": "kr-central", "name": "Korea Central", "country": "KR", "providers": 74, "gpus": 301, "utilization": 73.1, "lat": 37.57, "lon": 126.98 }
]

GET /v1/network/sku-mix

GPU count and share per SKU across the network. share is a fraction of all listed GPUs.

JSON
[
  { "sku": "rtx4090", "name": "GeForce RTX 4090", "gpus": 1459, "share": 0.26 },
  { "sku": "rtx3090", "name": "GeForce RTX 3090", "gpus": 954, "share": 0.17 },
  { "sku": "h100-sxm", "name": "NVIDIA H100 SXM", "gpus": 337, "share": 0.06 }
]

GET /v1/network/leaderboard

QueryDefaultDescription
limit20Number of providers.
JSON
[
  { "rank": 1, "address": "0x8b3f…2e41", "name": "kestrel-compute", "tier": "datacenter", "reliability": 98, "gpuHours30d": 41280, "jobs30d": 1873, "region": "eu-central" },
  { "rank": 2, "address": "0x1d9a…07c3", "name": "northbay-gpu", "tier": "pro", "reliability": 97, "gpuHours30d": 18944, "jobs30d": 2410, "region": "us-west" }
]

GET /v1/network/activity

A live feed of recent network events, newest first.

QueryDefaultDescription
limit30Number of events.
JSON
[
  { "t": "2026-11-04T09:14:52Z", "type": "job_completed", "sku": "rtx4090", "region": "kr-central", "amount": "284.338888", "durationS": 6604 },
  { "t": "2026-11-04T09:14:47Z", "type": "job_started", "sku": "h100-sxm", "region": "eu-central" },
  { "t": "2026-11-04T09:14:31Z", "type": "lock_added", "sku": "a100-80", "region": "us-east", "amount": "172.050000" },
  { "t": "2026-11-04T09:14:02Z", "type": "provider_joined", "sku": "rtx5090", "region": "pl-central" }
]

type is one of job_completed, job_started, provider_joined or lock_added. amount (MZG) and durationS are present only where they apply.