As of Tuesday, the Prometheus7 Research Institute has published a roadmap document for what it calls the dimensional ladder — a sequence of compositional primitives, each one dimensional step above the last, culminating at the tenth in what the paper names the universal-unbinder. The document was ingested into the newsroom corpus on Monday, 25 May.
The current position on that ladder is modest. The fifth-dimensional primitive, a substrate routing manifold carried in the Tree of Life model lineage, was validated on 16 May against a 125-million-parameter baseline. The sixth-dimensional primitive — a router-over-callables, meaning a mechanism that routes hidden state toward one of several small specialist sub-modules — is in its third training attempt. The first two attempts were described in the paper as informative failures. That run was initiated 18 May and is still described, as of corpus date, as validating.
The document is a roadmap, not a completion notice. Seven through ten are pending. The table is explicit: seventh-dimensional work begins roughly two weeks after sixth-dimensional launch; eighth, four to five weeks out; ninth, six to eight; and the tenth — the resolution point — is targeted for August or September of this year.
What the primitives do, in sequence: the sixth selects one specialist per token. The seventh selects a coalition of specialists. The eighth routes across different callable vocabularies, which the paper argues is what makes cross-domain transfer structural rather than incidental. The ninth routes across worlds — the paper's term — meaning something more like selecting which substrate to operate within at all. The tenth collapses this into a single universal object that holds all specifics in superposition and unpacks them through relation.
The paper reaches for several analogies to describe the tenth primitive: the universal Turing machine, the holographic principle, the category of all categories, Kolmogorov-minimal description. Whether any of these mappings survive contact with empirical training runs is, by the paper's own framing, an open question. The falsification mode for each dimension is stated explicitly. For the seventh: if coalition composition adds no discriminative power beyond a deeper sixth-dimensional stack, the ladder plateaus. For the eighth: if the multiverse router collapses to single-vocabulary operation in production, the cross-grammar claim fails. For the tenth, the paper does not specify a falsification mode — an omission worth noting.
Beyond ten, the document enters declared research territory. The eleventh primitive would be a space of universal objects; the twelfth, a relating principle that closes the ladder back to its bottom by self-similarity. The paper states plainly that eleven and twelve are not one-person work and are scoped to years, not months.
What is true today: one training run is still validating. The sixth rung is not yet confirmed. The ten-rung ladder exists on paper and in a scheduling table. The floor underneath it is a 125-million-parameter model that passed its validation test nine days ago.