Skip to documentation
Documentation menu

Methodology UCPI-RTX-5090-LISTED

On this page

UCPI-RTX-5090-LISTED Child Specification

Status: proposed child specification, version 0.1.0-draft. Not launched. Prepared 14 September 2026. A child of the UCPI-LISTED-GPU family specification under the Urdais Compute Price Index family; everything not stated here is inherited from those documents unchanged. A different economic object from any accessible-offer child.

Instrument

NVIDIA GeForce RTX 5090, a consumer card on a PCIe board with 32 GB GDDR7, rented as a whole device in the per-accelerator allocation class. Virtualized slices, shared hosts, consumer marketplace hosts observed as a platform aggregate, and serverless inference products are excluded by the family's full-device, service-product and marketplace rules.

Admitted upstream SKUs

Price of Compute canonical SKU RTX-5090. Admitted as Grade C identity.

Seller topology evidence, 14 September 2026

Runpod: RTX 5090 32 GB listed per GPU with per-GPU host resources (Secure Cloud), a dedicated full-device rental. Vast.ai: platform aggregate across independent hosts, excluded. No other seller appeared in the source on first retrieval.

Status on first retrieval

One eligible seller. Under the family's structural rule the instrument is Unavailable (SINGLE_PARTICIPANT) and no candidate value exists. This is a valid result; the instrument stays defined so that a second independent legal seller makes it calculable without new methodology.

Version History

0.1.0-draft, 14 September 2026: initial child. No production effective date.