A closer look
at the market.

Not yet. Requenta is in preparation: no GPU reserve is committed and trading has not opened. Join early access as a buyer or supplier.

When available capacity is plentiful relative to demand. Flexible workloads can use those quieter windows; there are no fixed off-peak hours or guaranteed discounts.

GPU-hours on dedicated nodes for a defined time window. Buyers run their workloads; sellers get paid for the compute they deliver.

Supply and demand meet in a separate market for each GPU specification and region. Supplier nodes become eligible when the price clears their net floor. The clearing methodology will be published before trading opens.

The planned index reflects cleared compute transactions for a GPU specification and region, with a 15-minute refresh. No price or history is published until real trades exist.

In the planned model, your accepted rate stays fixed for the hours you book. An extension needs a new quote and available capacity.

The first planned market is dedicated nodes of 8 H100 SXM GPUs in Europe. Exact locations and availability will be published once capacity is committed.

It is the time you have to accept a quote, not the duration of your run. Once accepted, the rate stays fixed for your booked hours. Final product terms will be confirmed before launch.

Delivered compute is planned to be metered per minute and consolidated into one invoice. The accepted rate will not change when the index moves.

Yes, with whole-node capacity in Europe, including unused reservations you can resell. You set an availability window and a minimum net price. Hardware verification comes before listing.

The planned model uses dedicated nodes, a tenant Kubernetes endpoint, verification before listing and a wipe at hand-back. Final operating guarantees will be published before paid access.

Refunds for undelivered compute are part of the planned model. Failure detection, service boundaries and the refund policy will be published before the first paid job.

A planned way to prepay for GPU-hours, with a target discount to spot. Non-expiry and resale are intended features; final terms will be confirmed before purchase.

A proposed monthly-volume agreement with a price ceiling, priority reserve access and compensation for flexibility during scarcity. Register your interest to discuss your workload.