Onton's Ontology 1 Claims 0.630 P@10 Against Google Shopping's 0.543 While Indexing 1% of the Catalog

MarkTechPost·medium signalXBlueskyLinkedInCopy link

Onton released Ontology 1 on August 2–3, 2026, a neurosymbolic product-search model evaluated on the released Subtext-Decor-90 benchmark by three multimodal judges (Claude Opus 4.8, Gemini 3.1 Pro, GPT-5.5): mean precision@10 of 0.630 versus 0.543 for Google Shopping and 0.469 for Amazon, winning 52 of 90 queries outright while indexing roughly 1% of either catalog. The architecture is an inspectable knowledge graph that decomposes vague predicates into checkable properties. Two caveats: judge agreement is modest (Krippendorff's alpha 0.465), and there are no open weights, no public API, and no pricing — partner access only.