Docs
Reservations

Request a GPU reservation

POST
/v1/reservations

Requests the GPUs in an offering for its term. Requires Idempotency-Key. Returns 202 with the reservation.

A request made with an agent credential starts awaiting_approval, with an approval whose url an owner of the org opens in our console to approve or decline it within 72 hours: hand that link to a person. Approved, it becomes requested; declined, or undecided by approval.expiresAt, it becomes declined. Any other request starts requested.

A requested reservation is invoiced for its total, in invoice: emailed to your org's billing email and due 10 days after the request or at the reservation's start, whichever is first. A card on file is never charged for it. Paying it confirms the reservation; one still unpaid at invoice.dueAt is declined. Once confirmed, a reservation cannot be withdrawn or refunded. You can withdraw a request with DELETE until it is confirmed, which voids its invoice.

When the window opens the reservation becomes active, and POST /v1/instances with reservation and a GPU size launches instances against it: any size of its GPU model with reservation in purchaseOptions, so its GPUs are split across sizes as you like. At the end of the term the platform deletes those instances.

Authorization

headerAuthorizationBearer <cc_...>

An API key of your org, sent as Authorization: Bearer cc_.... Keys start with cc_ and are otherwise opaque. Treat a key as a password: anyone holding it can act with its scopes until it expires or is revoked. Each operation's security requirement lists the scopes it needs; an empty list means any valid key. The MCP server at /mcp accepts agent keys and no other kind.

Scope: reservations:write

Header Parameters

Idempotency-Key*string

A unique key for this intent, such as a UUID. Reuse it for every retry of the same request. A missing key returns 400 idempotency_key_required. See Idempotency above.

Match^[!-~]+$
Length1 <= length <= 255
Api-Version*string

The API version the request is written against. This document describes 2026-10-15. A missing header returns 400 api_version_required, and a version the API does not serve returns 400 unsupported_api_version.

X-Correlation-ID?string

An id of your own for this request, or for a group of requests, to find them in our logs and yours. Logged with the request and returned unchanged in X-Correlation-ID. The request id itself is always the API's own, in X-Request-ID.

Match^[A-Za-z0-9._:-]+$
Length1 <= length <= 128
Org-Context?string

Admin keys only: the id of the org to act in. Required with an admin key, together with Access-Reason; with any other credential it returns 400 invalid_request. Every request that carries it is recorded in that org's operator-access log.

Match^org-[a-z0-9]([-a-z0-9]*[a-z0-9])?$
Length5 <= length <= 63
Access-Reason?string

Admin keys only: why the operator is acting in the org, such as a support ticket. Required with Org-Context. Recorded, verbatim, in the org's operator-access log.

Match^[^\x00-\x1F\x7F]+$
Length10 <= length <= 500

Request Body

application/json
  1. body

A request for the GPUs in an offering.

offering*ReservationOfferingId

The id of an offering from GET /v1/reservation-offerings. An id that has expired, been taken or is unknown returns 409 offering_expired; one that is not an offering id at all returns 400 invalid_request.

Match^ro-[a-z0-9]([-a-z0-9]*[a-z0-9])?$
Lengthlength <= 63
displayName?ReservationDisplayName

A name for people, shown with the reservation. It need not be unique and never identifies the reservation; use its id.

Match^[^\x00-\x1F\x7F]+$
Length1 <= length <= 63

Response Body

Accepted. A request made with an agent credential is awaiting_approval until an owner of the org decides on it in our console; any other is requested until its invoice is paid, which confirms it. A replay returns its current state.

application/json
  1. response

Whole GPUs held for your org over a term.

id*ReservationId

Opaque. Starts with r-; otherwise as instance ids.

Match^r-[a-z0-9]([-a-z0-9]*[a-z0-9])?$
Lengthlength <= 63
displayName?string

The display name the request gave. Absent if none.

status*ReservationStatus

Open set; accept values not listed. awaiting_approval: requested with an agent credential, and waiting for an owner of the org to approve it in our console; you can withdraw it. requested: waiting for its invoice to be paid; you can withdraw it, which voids the invoice. confirmed: paid for, and ours to honour; cannot be withdrawn. active: the window is open and GPU instances can launch. ended: the window closed and its instances were deleted. declined and withdrawn: closed before confirmation.

statusChangedAt*string
Formatdate-time
statusMessage?string

A note on the status, such as why a request was declined. Do not parse it.

approval?

Only for a reservation requested with an agent credential: the approval of an owner of the org that it waits for, or got.

offering*ReservationOfferingId

Opaque. Starts with ro-; otherwise as instance ids.

Match^ro-[a-z0-9]([-a-z0-9]*[a-z0-9])?$
Lengthlength <= 63
location*LocationId

A location id from GET /v1/locations, such as us-west-1.

Match^[a-z0-9]([-a-z0-9]*[a-z0-9])?$
Length1 <= length <= 63
tee*Tee

The trusted execution environment. Open set; accept values not listed.

gpu*

GPUs, whole and passed through to the guest. Each instance gets its count of the host's GPUs, which the host's device plugin picks, and only those GPUs are passed to it. With 8 GPUs the guest has the whole board and can attest its NVSwitch topology. With 1, 2 or 4, it shares the board's NVSwitches with other instances on the host, which the guest cannot verify, and separation between instances over NVLink is not yet verified.

start*string
Formatdate-time
end*string
Formatdate-time
term*Term

A whole number of days, as an ISO 8601 duration, from P7D to P365D.

Match^P([7-9]|[1-9][0-9]|[12][0-9]{2}|3[0-5][0-9]|36[0-5])D$
price*

The list price for the whole term, and what your org pays instead if it has a contract for the GPU model.

invoice?

The invoice for the reservation's total, once it is requested. Absent before then, and while billing is not set up.

gpusInUse*integer

GPUs that instances in the reservation use right now.

Range0 <= value
requestedBy*

Who requested the reservation, so that an owner deciding on an agent's request sees who asked. An agent's request carries approval.

createdAt*string
Formatdate-time
Example Request
curl -X POST 'https://api.cloud.confidential.ai/v1/reservations' \  -H 'Idempotency-Key: 0b9d6c1e-5f8a-4c2e-9a51-2f7e0c3d4b18' \  -H 'Api-Version: 2026-10-15' \  -H 'X-Correlation-ID: rollout-2026-10-22' \  -H 'Access-Reason: Support ticket 4821: customer asked why i-6d3f9h2k8mzr failed' \  -H 'Content-Type: application/json' \  -d '{  "offering": "ro-3hx8k2",  "displayName": "November training run"}'
Example Response
{  "id": "r-5mf2k9x3qp",  "displayName": "November training run",  "status": "requested",  "statusChangedAt": "2026-10-22T17:40:00Z",  "offering": "ro-3hx8k2",  "location": "us-west-1",  "tee": "tdx",  "gpu": {    "model": "b300",    "count": 8  },  "start": "2026-11-02T00:00:00Z",  "end": "2026-12-02T00:00:00Z",  "term": "P30D",  "price": {    "currency": "USD",    "perGpuHour": "7.00",    "total": "40320.00"  },  "invoice": {    "status": "open",    "url": "https://invoice.stripe.com/i/acct_1Example/test_YWNjdF8xRXhhbXBsZQ",    "dueAt": "2026-11-01T17:40:00Z"  },  "gpusInUse": 0,  "requestedBy": {    "userId": "user_01J8ZQ4X3N2M5K7P9R6T",    "agent": false  },  "createdAt": "2026-10-22T17:40:00Z"}