Onton Releases Ontology 1: A Neurosymbolic Search Model 2.7x More Accurate Than Top E-commerce Search Engines

Onton, a San Francisco-based search and discovery company, has released Ontology 1, a neurosymbolic model for complex, conversational, and multimodal product search. On a 90-query benchmark scored by three independent LLM judges, Ontology 1 reached a mean precision@10 of 0.630, compared to 0.543 for Google Shopping and 0.469 for Amazon, while indexing roughly 1% of their catalogs. The model is live at Onton.com, with partner access granted case by case; there is no public API or open checkpoint. Ontology 1 builds an explicit, inspectable world model rather than absorbing patterns into weights, and it reasons from objective properties like fiber, weave, and construction, flagging contradictory claims. It uses a custom graph database called Ograph, which Onton reports is significantly faster than SuiteSparse:GraphBLAS. The benchmark, Subtext-Decor-90, uses three multimodal judges: Claude Opus 4.8, Gemini 3.1 Pro, and GPT-5.5, with Krippendorff's alpha of 0.465 indicating moderate agreement, though all judges rank the engines in the same order. Ontology 1 wins 52 of 90 queries outright, but loses on functional-spec queries where Amazon's category metadata dominates, such as 'lamp that won't wake my partner if I read at 3am' (Onton 0.4, Amazon 0.9). The model currently only indexes home decor and furniture, but the methodology is said to generalize to other domains. Key takeaways include superior accuracy, a neurosymbolic architecture, and product-first deployment.