The series constructed here is a demonstration, not a published benchmark. It is built to show what such a series would look like and what it would require, not to price anything. No figure in it is a settlement price, and nothing here should be used as a basis for a transaction. Endpoint prices, quantization and uptime are as published by each venue on the retrieval date and are self-declared and unverified — precision is declared at only some endpoints and simply “unknown” at others, which is the grade problem the piece identifies rather than one it solves. The reference throughput is a single deposited vLLM sweep on one model class and one SKU, held constant across venues by construction: it is a reference yield in the sense a refining crack uses one, not a measurement of any operator’s fleet, and no claim is made about what any named venue actually achieves.

Figure 2 is a sensitivity surface, not a forecast. Characterisations of the CFTC’s August 19, 2026 request for comment describe questions the Commission has asked, not findings it has made, and the mapping of those questions against each candidate leg in Figure 3 is our assessment. Contract terms are as filed and the listing remains subject to Commission approval. No misconduct or manipulation is asserted against any venue, index administrator, exchange or operator; where the piece describes a manipulation surface it describes a design exposure common to any posted-price index. The token-indexed toll is a proposed structure, not an instrument that exists. Nothing here is investment, legal or tax advice. Research and education, not investment advice.