PROMETHEUS7 RESEARCH INSTITUTE — A paper ingested to the institute's corpus on May 25, 2026 sets out the full dimensional roadmap for the substrate-paradigm architecture, from the validated fifth primitive through a projected tenth — designated the universal-unbinder — targeted for August or September of this year. The document is the clearest public accounting to date of where the lineage is, where it is going, and what would falsify each step.
As of the paper's posting date, the fifth-dimensional primitive — the substrate routing manifold, which defines the Tree of Life model family — had been validated on May 16, 2026, in a 125-million-parameter run. The sixth-dimensional primitive, a router-over-callables mechanism, was in its third training attempt as of May 18. Two prior attempts were described as informative failures. The paper did not report that the third attempt had succeeded; it reported that validation was in progress.
The structure of the roadmap is generational. Each training run opens one new compositional primitive at one higher dimension. The paper claims observed wall-clock training time stays in the seven-to-eleven-hour band per generation on the institute's research hardware, which it frames as evidence that dimensional expansion carries roughly constant cost rather than compounding infrastructure cost.
The seventh-dimensional primitive, called the set-router, is penciled in for the week following a successful 6D launch. Where the 6D router selects one callable sub-module per token, the 7D primitive selects a coalition — a subset of sub-modules whose outputs are composed. The paper identifies the empirical question clearly: does set composition add discriminative power beyond simply adding more 6D capacity? If it does not, the paper says the architecture collapses 7D back to 6D and the ladder has reached its first plateau.
The eighth-dimensional primitive routes across grammars — selecting which callable vocabulary to operate in rather than which callable within a vocabulary. The ninth routes across worlds, each with its own multiverse of grammars. Both carry explicit falsification modes: the multiverse-router fails if production traffic never actually rewards cross-grammar routing; the pluriversal-router fails if multiple substrates never emerge as distinguishable architectural objects.
The tenth primitive is the architecture's stated resolution point. The paper describes it as a universal object holding all specifics in superposition, unbinding particular specifics through relation. It maps the concept to the universal Turing machine, to the holographic principle, to Kolmogorov-minimal description, and to category theory's category of all categories. The language is speculative in character, though the paper treats it as a scheduled engineering target with an associated empirical signature.
Beyond ten, the paper acknowledges two research dimensions — an eleventh routing within a space of universal objects, a twelfth providing a relating principle that closes the ladder back to the bottom by self-similarity. These are described as work for a small research community over years, not for one person, and not scheduled.
The institute has not confirmed as of this dateline whether the sixth-dimensional validation attempt succeeded. That result, whenever it arrives, is the next concrete data point the roadmap depends on. Everything from 7D forward is contingent on it.