GetDeploying API

Stop wrestling with fragmented pricing models. Our API aggregates and normalizes GPU market data from 91 cloud providers into a unified, queryable schema. We provide the market intelligence you need to:

• Benchmark costs across major hyperscalers and specialized GPU clouds
• Programmatically filter by GPU model, VRAM, provider, billing type, and price range
• Track pricing trends with most listings re-checked from every 15 minutes to once a day
• Deploy cost-aware infrastructure using structured, standardized datasets

Authentication

Getting an API key

Start with a free trial: sign up with a work email for 100 requests, no card needed. For ongoing access, see the plans.

Using your API key

Authenticate by including your API key in the Authorization header as a Bearer token:

curl -H "Authorization: Bearer YOUR_API_KEY" \
  "https://getdeploying.com/api/gpu-offerings"

Rate limits

The daily limit is shared by every API key on your account. The API includes rate limit information in response headers. For example:

X-RateLimit-Limit: 3000
X-RateLimit-Remaining: 2847
X-RateLimit-Reset: 43200

Here's what each header indicates:

• X-RateLimit-Limit: Maximum requests allowed per period
• X-RateLimit-Remaining: Requests remaining in current period
• X-RateLimit-Reset: Seconds until the rate limit resets

Rate limit exceeded

When you exceed your rate limit, you'll receive a 429 Too Many Requests response:

{
  "error": "Rate limit exceeded. Try again in 43200 seconds."
}

Trial requests

Trial responses carry two more headers:

X-Trial-Limit: 100
X-Trial-Remaining: 87

• X-Trial-Limit: Requests the trial includes
• X-Trial-Remaining: Requests left in the trial

Once none are left, the API answers 402 Payment Required:

{
  "error": "Trial quota of 100 requests used. Contact [email protected] to continue."
}

Error codes

Error response format

Error responses include a JSON object with an error field:

{
  "error": "Invalid token"
}

List GPU offerings

Request

GET /api/gpu-offerings

Returns a paginated list of GPU offerings from cloud providers, with detailed specifications and pricing information.

Query parameters

Response format

The API returns paginated JSON with an array of GPU offerings:

{
  "page": 1,
  "page_size": 10,
  "page_count": 4,
  "total": 38,
  "data": [
    {
      "id": "aws-a1b2c3d4e5f6a7b8",
      "external_id": "p5.48xlarge",
      "provider": {
        "id": "aws",
        "name": "Amazon Web Services",
        "website": "https://aws.amazon.com/",
        "country": "US"
      },
      "configuration": {
        "gpu_model": "nvidia-h100",
        "gpu_count": 8,
        "vram_per_gpu_gb": 80,
        "vram_total_gb": 640,
        "gpu_mem_bandwidth_gbps": 3350,
        "supported_precisions": ["FP8", "FP16", "BF16", "TF32", "FP32", "FP64", "INT8"],
        "interconnect_bandwidth_gbps": 900,
        "interconnect": "SXM",
        "cpu_cores": 192,
        "system_ram_gb": 2048,
        "disk_storage_gb": null
      },
      "pricing": {
        "currency": "USD",
        "billing_type": "ON_DEMAND",
        "hourly": 98.32,
        "hourly_per_gpu": 12.29,
        "monthly": 70790.4,
        "note": "On-demand pricing"
      },
      "status": {
        "availability": "AVAILABLE",
        "note": "Instance available at time of verification.",
        "last_verified": "2026-01-15T12:00:00Z"
      }
    }
  ]
}

Field descriptions

Top-level fields:

Provider (provider):

Configuration (configuration):

Pricing (pricing):

Status (status):

AVAILABLE means the provider listed it as available at last_verified. It is not a stock guarantee. WAITLIST means it is available on request, or can be ordered now and delivered later (note says when).

Pagination:

Examples

# Get all H100 offerings
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-h100"

# Filter by provider and billing type
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?provider=aws&billing_type=ON_DEMAND"

# Query multiple GPU models and providers
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-h100,nvidia-a100&provider=aws,microsoft-azure"

# Find cheapest H100 offerings under $30/hour, sorted by price per GPU
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-h100&max_price=30&sort=price_per_gpu_hour"

# Get large H100 instances (8+ GPUs) sorted by total price
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-h100&min_gpu_count=8&sort=price_hour"

# Sort by descending GPU count (largest instances first)
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-h100&sort=-gpu_count"

# Get all A100 offerings from European providers
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?gpu_model=nvidia-a100&provider_country=EU"

# Filter by minimum VRAM per GPU
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?min_vram=80&sort=price_per_gpu_hour"

List GPU price history

GET /api/gpu-price-history

Returns pricing snapshots for GPU models. By default, returns aggregate statistics across all providers. Use the provider parameter to get price history for a specific cloud provider.

By default this endpoint returns the weekly series: one snapshot per week, taken on a Monday. Pass interval=daily to get every daily snapshot instead.

Daily snapshots start on 30 July 2026. Before that date, only weekly records are available, so interval=daily returns the same Monday snapshots as interval=weekly for earlier dates.

How snapshots are built

Each daily snapshot is one reading of the offerings at 10:00 UTC, not a summary of the day. It counts offerings that are still listed, have a price and are not UNAVAILABLE. A day with no such offering has no row.

• If a provider's prices could not be collected, its last collected prices are used.
• Each reservation_months value is a separate RESERVATION series.
• interval=weekly returns each week's Monday snapshot. A week with no Monday row is missing; interval=daily returns its other days.
• A provider whose GPU listings were withdrawn returns no history.

Query parameters

Response format

{
  "page": 1,
  "page_size": 10,
  "page_count": 3,
  "total": 24,
  "data": [
    {
      "date": "2026-01-05",
      "gpu_model": "nvidia-h100",
      "provider": null,
      "billing_type": "ON_DEMAND",
      "reservation_months": null,
      "currency": "USD",
      "min_price": "1.79",
      "max_price": "12.29",
      "avg_price": "5.44",
      "median_price": "4.76",
      "provider_median_price": "4.91",
      "offering_count": 42,
      "provider_count": 18
    }
  ]
}

Field descriptions

Examples

# Market-wide H100 on-demand price history (aggregated across all providers)
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?gpu_model=nvidia-h100&billing_type=ON_DEMAND"

# Same query, but scoped to a single provider
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?provider=runpod&gpu_model=nvidia-h100&billing_type=ON_DEMAND"

# Compare two providers' H100 pricing (make two calls)
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?provider=aws&gpu_model=nvidia-h100"
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?provider=runpod&gpu_model=nvidia-h100"

# A100 spot price history
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?gpu_model=nvidia-a100&billing_type=SPOT"

# Filter by date range
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?gpu_model=nvidia-h100&billing_type=ON_DEMAND&date_start=2026-01-01&date_end=2026-03-31"

# Paginate through a large result set (update page param as needed)
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-price-history?provider=aws&page_size=100&page=1"

List GPU models

GET /api/gpus

Returns all GPU models tracked in the database. Useful for discovering valid gpu_model slugs to use as filters, and for the specs behind an offering.

[
  {
    "id": "nvidia-h100",
    "name": "Nvidia H100",
    "vendor": "NVIDIA",
    "market_class": "DATACENTER",
    "performance_segment": "HIGH_PERFORMANCE",
    "release_date": "2022-09-20",
    "vram_gb": 80,
    "memory_type": "HBM3",
    "mem_bandwidth_gbps": 3350,
    "supported_precisions": ["FP8", "FP16", "BF16", "TF32", "FP32", "FP64", "INT8"],
    "interconnect": "SXM",
    "interconnect_bandwidth_gbps": 900,
    "variants": [
      {"interconnect": "SXM", "vram_gb": 80, "mem_bandwidth_gbps": 3350, "interconnect_bandwidth_gbps": 900},
      {"interconnect": "PCIe", "vram_gb": 80, "mem_bandwidth_gbps": 2000, "interconnect_bandwidth_gbps": 600},
      {"interconnect": "NVL", "vram_gb": 94, "mem_bandwidth_gbps": 3900, "interconnect_bandwidth_gbps": 600}
    ],
    "datasheet_url": "https://resources.nvidia.com/en-us-gpu-resources/h100-datasheet-24306"
  }
]

Field notes:

• market_class is the product line (DATACENTER, WORKSTATION, CONSUMER) and performance_segment the speed tier (HIGH_PERFORMANCE, MID_RANGE, BUDGET).
• The top-level spec fields describe one card, the one interconnect names. An H100 SXM runs at 3,350 GB/s, the PCIe card at 2,000.
• variants lists every form factor the model ships in, the one the top-level fields describe first. It is empty for a model sold in a single form factor, which is most of them. Each entry states its card in full, and a null interconnect_bandwidth_gbps means that card has no interconnect fabric rather than an unknown figure.
• Per-offering specs are resolved the same way: an offering's configuration.gpu_mem_bandwidth_gbps is its own form factor's figure, not the headline.
• Fields with nothing recorded are null.

List providers

GET /api/providers

Returns all active cloud providers. Useful for discovering valid provider slugs to use as filters.

[
  {
    "id": "aws",
    "name": "Amazon Web Services",
    "website": "https://aws.amazon.com/",
    "country": "US"
  }
]

Best practices

Caching

Pricing is collected on a per-provider schedule, from every 15 minutes to once a day. No provider is collected more often than every 15 minutes, so a shorter cache cannot return anything newer.

If you're sending a high volume of requests, we recommend:
- Cache responses for at least 15 minutes
- Store historical data locally for trend analysis

Pagination

The API returns paginated results with 10 items per page by default. Use the page and page_size parameters to navigate:

# Get page 2 with 100 results per page
curl -H "Authorization: Bearer YOUR_TOKEN" \
  "https://getdeploying.com/api/gpu-offerings?page=2&page_size=100"

Filtering efficiently

For best performance:
- Use specific filters (gpu_model, provider) rather than fetching all data
- Filter on the API side rather than locally to reduce bandwidth
- Combine multiple filters to narrow results

Supported GPU models

The API includes pricing for 110 GPU models:

View detailed GPU specifications →

Supported providers

The API covers 91 cloud providers:

View all providers with details →

Data latency & accuracy

• Check frequency: Most offerings are re-checked every 15 minutes or hourly, and the rest once a day. Cadence follows volatility, not catalogue size: providers whose rates and capacity move through the day sit on the fastest tier, while a provider with a stable published price list is checked daily. A small number of providers whose prices rarely change are maintained by hand. Each offering carries its own status.last_verified timestamp.
• Non-binding: All prices returned by the API are estimates for informational purposes only and do not constitute a binding offer to sell compute.
• Final billing: The final cost is determined strictly by the provider at the moment of provisioning. We strongly recommend performing a dry-run or checking the provider's console before launching high-volume workloads.
• Some providers may have regional price variations and other pricing factors not reflected in our data.
• Contact us if you notice any discrepancies.

Legal & usage

Use of the API is subject to our Terms and Conditions.

Support

You can reach us anytime at [email protected].