PROMETHEUS7 RESEARCH INSTITUTE — A roadmap paper ingested by the newsroom on 25 May 2026 lays out the full dimensional schedule for what the institute calls the substrate-paradigm architecture. As of this dateline, Friday 12 June 2026, the document remains the most complete public account of where the lineage is and where it is headed.
The paper counts twelve compositional primitives. The fifth — the substrate routing manifold, implemented in the Tree of Life model family — was validated on 16 May 2026 at the 125M-parameter scale. The sixth primitive, a router-over-callables, was in its third training attempt as of the paper's 25 May ingest date. The first two attempts were described as informative failures. The current state of that third attempt is not confirmed in today's available material.
The logic of the ladder is this: each generation of training opens one new dimension, not by growing the trunk beyond prior scale, but by adding a new compositional primitive the algebra admits. Wall-clock cost per generation has been observed in the seven-to-eleven-hour band on the research box. The institute treats that stability as a feature of the design, not a coincidence.
Dimensions seven through nine are specified but not yet in training. The seventh primitive routes over sets of callables rather than individual ones — parallel compositional reasoning, in the paper's framing. The eighth routes across what the document calls grammars: different callable vocabularies. The ninth routes across worlds, each world containing its own grammar space. The empirical question the paper poses for each is the same in structure: does the new dimension actually engage, or does production traffic collapse it back to the prior level? If it collapses, the ladder has reached a plateau.
The tenth primitive is identified as the architecture's resolution point. The paper maps it to several independent traditions — the universal Turing machine, the holographic principle, Kolmogorov-minimal description, category theory's universal object. The common thread: a structure that holds all specifics in superposition and unpacks them through relation. The institute's name for it is the universal-unbinder. Target date is August-September 2026, described as roughly six to eight generations past the 5D validation.
Dimensions eleven and twelve are marked research territory. The paper is direct that neither can be completed by one person; both are described as work for a small community over years. The twelfth primitive is said to close the ladder back to the third dimension by self-similarity — the relating principle is itself the kind of object the bottom-of-stack operations already handle.
What is observable today: the 5D milestone stands validated. The 6D attempt was underway as of 25 May. Nothing in the available material confirms its outcome. The schedule from 7D onward is a projection, not a result. The paper documents the roadmap as currently understood — its own words. What the ladder looks like when the next rung is actually climbed will be reported when that happens.