Pricing API

#Overview

Base URL: https://cryptgpu.com/api/v1. Every endpoint answers GET requests with JSON. Prices are in US dollars per month, prepaid; meta.market_checked is the date of the market check the prices come from.

#List GPUs

GET/api/v1/gpusOpen

Returns every GPU model with its specifications, server sizes, price per GPU per month, resources per GPU and market reference.

curl https://cryptgpu.com/api/v1/gpus

Response (first item shown):

{
    "meta": {
        "api_version": "v1",
        "currency": "USD",
        "prices": "per month, prepaid",
        "market_checked": "2026-09-23"
    },
    "data": [
        {
            "id": "b300",
            "name": "NVIDIA B300",
            "vendor": "NVIDIA",
            "architecture": "Blackwell Ultra",
            "family": "flagship",
            "memory_gb": 288,
            "memory_type": "HBM3e",
            "memory_bandwidth": "8 TB/s",
            "interconnect": "NVLink 5 · 1.8 TB/s",
            "precisions": ["FP4", "FP8", "BF16"],
            "server_sizes": [1, 2, 4, 8],
            "price_per_gpu_month": 4019,
            "resources_per_gpu": {
                "vcpu": 32,
                "ram_gb": 384,
                "nvme_tb": 3.84
            },
            "market": {
                "median_per_gpu_month": 5745,
                "lowest_per_gpu_month": 4818,
                "highest_per_gpu_month": 6278,
                "quotes": 8,
                "indicative": false
            },
            "savings_vs_median_pct": 30,
            "url": "https://cryptgpu.com/gpus/b300/"
        }
    ]
}

Fields

#Get a GPU

GET/api/v1/gpus/{id}Open

Returns one GPU, with the same fields as the list. An unknown id returns 404 with the list of valid ids.

curl https://cryptgpu.com/api/v1/gpus/h200

{
    "meta": {
        "api_version": "v1",
        "currency": "USD",
        "prices": "per month, prepaid",
        "market_checked": "2026-09-23"
    },
    "data": {
        "id": "h200",
        "name": "NVIDIA H200",
        "vendor": "NVIDIA",
        "architecture": "Hopper",
        "family": "flagship",
        "memory_gb": 141,
        "memory_type": "HBM3e",
        "memory_bandwidth": "4.8 TB/s",
        "interconnect": "NVLink 4 · 900 GB/s",
        "precisions": ["FP8", "BF16", "FP64"],
        "server_sizes": [1, 2, 4, 8],
        "price_per_gpu_month": 2279,
        "resources_per_gpu": {
            "vcpu": 24,
            "ram_gb": 256,
            "nvme_tb": 3.84
        },
        "market": {
            "median_per_gpu_month": 3259,
            "lowest_per_gpu_month": 2548,
            "highest_per_gpu_month": 4606,
            "quotes": 12,
            "indicative": false
        },
        "savings_vs_median_pct": 30,
        "url": "https://cryptgpu.com/gpus/h200/"
    }
}

#Quote a server

GET/api/v1/quote?gpu={id}&n={gpus}Open

Returns the monthly price and resources of a server.

curl "https://cryptgpu.com/api/v1/quote?gpu=h200&n=4"

{
    "meta": {
        "api_version": "v1",
        "currency": "USD",
        "prices": "per month, prepaid",
        "market_checked": "2026-09-23"
    },
    "data": {
        "gpu": "h200",
        "gpu_name": "NVIDIA H200",
        "gpus": 4,
        "currency": "USD",
        "price_month": 9116,
        "price_per_gpu_month": 2279,
        "hourly_equivalent": 12.49,
        "market_median_month": 13036,
        "savings_month": 3920,
        "savings_year": 47040,
        "server": {
            "gpu_memory_gb": 564,
            "vcpu": 96,
            "ram_gb": 1024,
            "nvme_tb": 15.36,
            "interconnect": "NVLink 4 · 900 GB/s"
        },
        "billing": "monthly, prepaid",
        "configure_url": "https://cryptgpu.com/configure/?gpu=h200&n=4"
    }
}

In JavaScript:

const res = await fetch('https://cryptgpu.com/api/v1/quote?gpu=h200&n=4');
const { data } = await res.json();
console.log(`${data.gpus}× ${data.gpu_name}: $${data.price_month}/month`);

#Errors

Errors return a JSON body with an error object and the matching HTTP status.

{
    "error": {
        "status": 400,
        "code": "invalid_size",
        "message": "The NVIDIA RTX 5090 comes in servers of 1, 2 or 4 GPUs.",
        "valid_sizes": [1, 2, 4]
    }
}

#Rate limits and caching

• Up to 120 requests per minute per IP address. Every response carries X-RateLimit-Limit and X-RateLimit-Remaining.
• Successful responses can be cached for 5 minutes (Cache-Control: public, max-age=300) and carry an ETag: send it back in If-None-Match to get a 304 Not Modified.
• CORS is open (Access-Control-Allow-Origin: *): you can call the API from a browser.
• Prices change only when we publish a new market check. Polling more than once an hour brings nothing new.

#Changelog

v1 — 23 September 2026. First release: /gpus, /gpus/{id} and /quote. New fields may be added to v1 responses; existing fields will not change meaning within v1.