Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ornn.com/llms.txt

Use this file to discover all available pages before exploring further.

Ornn uses a bid-based pricing model rather than fixed spot pricing. When you want to reserve GPU capacity, you submit a bid that states how much you’re willing to pay per GPU-hour. The Ornn team reviews incoming bids and accepts the ones that best match available supply. This model gives you control over your compute budget and lets Ornn match capacity to tenants efficiently.

How bidding works

When you complete the reservation form at /reserve/[deploymentId], you submit a bid. The bid records your desired GPU count, the date range you want, and the price per GPU-hour you’re offering. Ornn holds your bid in a pending state while it’s under review. The Ornn team evaluates bids and either accepts or rejects them. An accepted bid creates a confirmed reservation and triggers the checkout and invoicing flow. A rejected bid closes without creating a reservation.

Bid fields

When you submit a reservation, the following fields make up your bid:
deployment_id
UUID
required
The ID of the inventory deployment (GPU cluster) you want to reserve. This is set automatically from the listing you selected.
gpu_count
integer
required
The number of GPUs you want to reserve. Must be at least 1.
min_gpu_count
integer
The minimum number of GPUs you would accept if the full gpu_count is not available. If omitted, the full count is required. Must not exceed gpu_count.
start_date
date (YYYY-MM-DD)
The date you want the reservation to begin. Set from the calendar selector on the reservation page.
end_date
date (YYYY-MM-DD)
The date you want the reservation to end. Must be on or after start_date.
bid_price_per_gpu_hour
number
required
Your offered price in USD per GPU-hour. Must be greater than 0.
min_gpu_count lets you express flexibility on the accepted quantity. For example, if you set gpu_count to 16 and min_gpu_count to 8, you’re telling Ornn that you’ll accept anywhere from 8 to 16 GPUs. The accepted bid response tells you the exact GPU count that was confirmed.

Bid response fields

Once a bid is reviewed, you can retrieve it from your portfolio at /portfolio/bids. The following fields are returned:
id
UUID
The unique identifier for this bid.
status
string
The current bid status. See the status table below.
bid_price_per_gpu_hour
number
The price per GPU-hour you submitted.
accepted_at
datetime
The timestamp when the bid was accepted. null if not yet accepted.
accepted_gpu_count
integer
The number of GPUs confirmed in the accepted bid. May differ from gpu_count if partial acceptance was applied. null if not yet accepted.

Bid statuses

StatusMeaning
pendingSubmitted and awaiting review
acceptedApproved by the Ornn team; a confirmed reservation has been created
rejectedThe bid was not accepted; no reservation was created
cancelledYou cancelled the bid before it was reviewed

Gauging market price with the bid summary

Before submitting a bid, you can get a sense of competitive pricing using the bid summary data surfaced on the reservation form. Ornn exposes two views of historical bid activity for each inventory item: Monthly bid summary — Returns the highest bid price per GPU-hour for each calendar month, across all bids on that deployment. This gives you a long-range view of demand and pricing trends. Daily bid summary — Returns the highest bid price per GPU-hour for each day within a specific date range. Use this for finer-grained pricing insight when you’re targeting a specific window. The bid summary chart on the /reserve/[deploymentId] page draws from these data points. A higher max bid price in a given period indicates stronger demand for that deployment.
If the chart shows consistently high max bid prices in the period you want, consider bidding at or above that level to improve your acceptance chances.

Updating or cancelling a bid

You can update or cancel a pending bid from /portfolio/bids before it is accepted or rejected. To update a bid, navigate to the bid detail page at /portfolio/bids/[bidId] and edit the GPU count, date range, or price. All fields — min_gpu_count, gpu_count, start_date, end_date, and bid_price_per_gpu_hour — must be submitted together when updating. To cancel a bid, click Cancel bid on the bid detail page. Cancellation sets the bid status to cancelled and closes the bid permanently. Cancelled bids cannot be reopened.
You can only update or cancel a bid while it has pending status. Once a bid is accepted, rejected, or cancelled, no further changes are possible.

What’s next

Completing checkout

After your bid is accepted, complete checkout to finalize billing and confirm your reservation.

Managing your bids

View all your bids, track statuses, and manage pending offers from your portfolio.