Skip to main content
POST
/
get_offers
Get Offers
curl --request POST \
  --url https://supply-api.compute-desk.com/get_offers \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/yaml' \
  --data '
"vendor: nebius\nregion: EU\n\ncompute:\n  gpu_type: nvidia-l40s-48gb\n  gpu_count: 1\n  max_price_per_gpu_hour: 5\n  contract_type: ondemand\n"
'
{
  "offers": [
    {
      "offer_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
      "gpu_type": "<string>",
      "gpu_family": "<string>",
      "gpu_count": 123,
      "vendor": "<string>",
      "region": "<string>",
      "gpu_price_per_hour": 123,
      "boot_disk_in_gb": 123,
      "created_at": "2023-11-07T05:31:56Z",
      "vendor_region": "<string>",
      "contract_type": "<string>",
      "type": "vm",
      "vcpu_count": 123,
      "memory_in_gb": 123,
      "storage_details": [
        {
          "name": "<string>",
          "size_gb": 123,
          "type": "<string>",
          "storage_class": "<string>",
          "vendor_class": "<string>",
          "vendor_region": "<string>",
          "price_per_gb_month": 123,
          "monthly_cost": 123,
          "hourly_cost": 123,
          "persistent": false
        }
      ],
      "storage_price_per_hour": 123,
      "total_price_per_hour": 123,
      "expires_at": "2023-11-07T05:31:56Z",
      "deployment_type": "<string>",
      "node_count": 123,
      "gpus_per_node": 123,
      "infiniband_bandwidth_gbps": 123,
      "cluster_storage_gb": 123
    }
  ],
  "total_count": 123,
  "filters_applied": {}
}
Query real-time GPU offers across all vendors. This is the primary entry point for finding available compute.
This endpoint accepts a YAML request body, not JSON. Set Content-Type: application/yaml.

​
Request format

vendor: nebius              # Optional: filter by vendor
region: EU                  # Optional: US, EU, APAC, Other
compute:
  gpu_type: H100            # Family name or canonical name
  gpu_count: 1              # Number of GPUs
  max_price_per_gpu_hour: 5 # Price cap in USD
  contract_type: ondemand   # ondemand or spot
network:                    # Optional
  supports_cilium: true     # Filter to Cilium-capable vendors

​
Key fields

FieldRequiredDescription
compute.gpu_typeYesGPU family (e.g., H100) or canonical name (e.g., nvidia-h100-sxm5-80gb)
compute.gpu_countNoNumber of GPUs (default: 1)
compute.max_price_per_gpu_hourNoMax price in USD per GPU per hour
compute.contract_typeNoondemand or spot
vendorNoFilter to a specific vendor (e.g., nebius, hyperstack, scaleway)
regionNoFilter to a region group: US, EU, APAC, Other

​
Rate limiting

This endpoint is rate-limited to 5 requests per minute per client. Exceeding the limit returns 429 Too Many Requests.

​
Offer expiration

Offers expire after 30 minutes. Use the offer_id to procure before expiration.

​
Filtering by vendor

Pass the vendor name directly: 
vendor: massedcompute
compute:
  gpu_type: A100
  gpu_count: 1

Authorizations

​
Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/yaml

The body is of type string.

Response

Successful Response

Response containing available offers.

​
offers
GetOffersSummary · object[]
required
​
total_count
integer
required
​
filters_applied
Filters Applied · object
required

Filters that were applied to the query