Trade is the single platform AI Model Providers use to manage orders and run them. Place, cancel and reprice resting asks against live book depth, let algorithmic yield work the book autonomously, and clear and settle every order with an auditable record — while a micro-batching layer smooths matched job streams into your fleet to keep every node near optimal utilisation.
Trade gives your algorithms direct access to the order book — through an embeddable SDK, a graphical UI and an autonomous trading-agent layer. Post resting asks, adjust price and depth, and cancel or reprice in seconds as fleet utilisation shifts, so idle GPU blocks clear without eroding your premium direct pricing.
A micro-batch controller paces every matched job stream into your fleet — absorbing demand spikes and turning idle headroom into higher effective yield.
Keld handles clearing and settlement end-to-end so you can focus on running great infrastructure. Every order is logged with a transparent, independently auditable record.
Sell idle compute into live demand instead of letting it sit. A channel alongside your existing ones.
Trade unifies the financial plane — order management, yield and settlement — with the data plane that micro-batches matched jobs onto your fleet.
List on Keld alongside any other channel. Named-model guarantees enforced end-to-end.
We're onboarding AI Model Providers through the private beta.