Pricing API.

PRICING API · THEAI LTD · INNOVATION ONE, DIFC, DUBAI, UAE · LAST UPDATED SEPTEMBER 11, 2026

CHANGELOG · SEP 11 — B300 AVAILABILITY IS NOW "AVAILABLE" (ORDERABLE), DELIVERY WINDOW STILL IN THE WRITTEN QUOTE; SEP 8 — AVAILABLE:TRUE ON BOTH SKUS; SEP 7 — H200 OFFERING; SEP 4 — UPDATED_AT, DELIVERY, ON_DEMAND, QUOTE/RATE-CARD URLS, COUNTRY FIELDS

A public, read-only JSON feed of our GPU pricing for aggregators and marketplaces. No API key, no authentication — a plain GET. The response is generated straight from our live rate card, cached for one hour, and served with Access-Control-Allow-Origin: * so it can be fetched from anywhere. For the authenticated console API (launch, billing, tokens) see Docs.

ENDPOINT

Method	Path	Auth
`GET`	`/api/v1/pricing`	none (public)

curl -s https://cloud.theai.com/api/v1/pricing | jq

RESPONSE FIELDS

Field	Type	Notes
`provider`	string	Always `"TheAI Cloud"`.
`operator_country`	string	`"AE"` — the operator, TheAI Ltd (DIFC, Dubai). Hardware location is `datacenter_country`.
`currency`	string	ISO 4217, `"USD"`.
`unit`	string	Price basis — `"per_gpu_hour"`.
`updated_at`	string	ISO-8601 date of the last rate-card change. Stable between changes — it is not the request time, so it is safe to show as "last updated".
`quote_url`	string	Where to request a written quote (site, delivery window, spec, rate — within 24 hours).
`rate_card_url`	string	Human-readable rate card and comparison table.
`offerings[]`	array	One entry per GPU SKU.
`offerings[].gpu`	string	GPU model — `"NVIDIA B300"` (reserved) or `"NVIDIA H200"` (available now).
`offerings[].sku`	string	Stable machine id of the configuration.
`offerings[].gpus_per_node`	number	GPUs in the smallest bookable unit (a node = 8 for both SKUs).
`offerings[].min_gpus` · `max_gpus`	number	Bookable range (`max_gpus` only where capped).
`offerings[].interconnect`	string	Fabric description.
`offerings[].region` · `country` · `datacenter_country`	string \| null	B300: `us` · `US` · `US` — the data center is in the United States (Tier III). Use `datacenter_country` for location filters. H200: `us` · `US` · `US` — US East.
`offerings[].availability`	string	`available` on every SKU we can take an order for today — both B300 and H200. It says the capacity is orderable, not that a node is powered on: use `available_from` to tell the two apart (H200 carries a date, B300 carries `null` because its delivery window is named in the written quote).
`offerings[].available`	boolean	`true` on every SKU we can take an order for today. Same meaning as `availability` — kept for readers that expect a boolean.
`offerings[].available_from`	string \| null	B300: `null` — we do not publish an availability date; the delivery window is confirmed per agreement, do not render a launch date or a waitlist. H200: ISO-8601 date since which nodes are available.
`offerings[].min_term_months`	number	Shortest term offered: 1 (B300), 3 (H200).
`offerings[].delivery`	string	`"confirmed in written quote"` — how the delivery window is set.
`prices.on_demand`	object \| null	B300: `{ status: "planned", price: null }` — on-demand is planned (date TBA); no on-demand price is published, so please do not list one. Becomes `{ status: "live", price, billing: "per_minute" }` once it opens. H200: `null` — not offered on-demand.
`prices.reserved[]`	array	`{ term_months, price }`, ascending by term. $/GPU·hour at the committed term (B300: 1, 3, 6, 12, 24 months; H200: 3, 6, 12); whole 8-GPU nodes.

EXAMPLE RESPONSE

{
  "provider": "TheAI Cloud",
  "operator_country": "AE",
  "currency": "USD",
  "unit": "per_gpu_hour",
  "updated_at": "2026-09-07T00:00:00.000Z",
  "quote_url": "https://cloud.theai.com/#rack",
  "rate_card_url": "https://cloud.theai.com/#pricing",
  "offerings": [
    {
      "gpu": "NVIDIA B300",
      "sku": "B300_SXM6_288GB",
      "gpus_per_node": 8,
      "min_gpus": 8,
      "interconnect": "NVLink 5 intra-node, 400G InfiniBand NDR inter-node",
      "region": "us",
      "country": "US",
      "datacenter_country": "US",
      "availability": "available",
      "available": true,
      "available_from": null,
      "delivery": "confirmed in written quote",
      "min_term_months": 1,
      "prices": {
        "on_demand": {
          "status": "planned",
          "price": null
        },
        "reserved": [
          {
            "term_months": 1,
            "price": 6.75
          },
          {
            "term_months": 3,
            "price": 6.55
          },
          {
            "term_months": 6,
            "price": 5.7
          },
          {
            "term_months": 12,
            "price": 5.3
          },
          {
            "term_months": 24,
            "price": 4.9
          }
        ]
      }
    },
    {
      "gpu": "NVIDIA H200",
      "sku": "H200_SXM5_141GB",
      "gpus_per_node": 8,
      "min_gpus": 8,
      "max_gpus": 48,
      "interconnect": "NVLink 4 intra-node",
      "region": "us",
      "country": "US",
      "datacenter_country": "US",
      "availability": "available",
      "available": true,
      "available_from": "2026-09-07T00:00:00.000Z",
      "delivery": "confirmed in written quote",
      "min_term_months": 3,
      "prices": {
        "on_demand": null,
        "reserved": [
          {
            "term_months": 3,
            "price": 3.7
          },
          {
            "term_months": 6,
            "price": 3.45
          },
          {
            "term_months": 12,
            "price": 3.2
          }
        ]
      }
    }
  ]
}

Prices are the single source of truth for our rate card — the same numbers shown on Pricing. Reserved terms are $/GPU·hour when committing for that many months, billed monthly per the capacity agreement after acceptance; on-demand is planned (date TBA) and has no published price. Questions or bulk feeds: cloud@theai.com.