FrontierAugust 3, 2026via MarkTechPost
Onton Releases Ontology 1: A Neurosymbolic Search Model That is 2.7x More Accurate than the World’s Best E-commerce Search Engines
Why it matters
A production search model claims significant accuracy gains over incumbents using a hybrid neurosymbolic approach. For practitioners building search/discovery systems, this is a capability benchmark worth stress-testing; for enthusiasts tracking model architectures, neurosymbolic reasoning is a growing frontier pattern beyond pure scaling.
Key signals
- Ontology 1: mean precision@10 of 0.630
- Google Shopping: 0.543 precision@10
- Amazon: 0.469 precision@10
- Benchmark: 90-query evaluation, scored by three independent LLM judges
- Onton indexed ~1% of competitor data to achieve results
- Model approach: neurosymbolic search, conversational + multimodal
- Company: Onton, San Francisco-based search and discovery startup
- Ontology 1: neurosymbolic model for product search
- Ontology 1 precision@10: 0.630
- Google Shopping precision@10: 0.543
- Amazon precision@10: 0.469
- Ontology 1 indexed ~1% of competitor corpus
- 2.7x improvement claim (0.630 vs 0.469 Amazon baseline)
- Company: Onton, San Francisco
- Model type: neurosymbolic, conversational, multimodal
The hook
0.630 precision@10. Onton's neurosymbolic search model just beat Google Shopping and Amazon on a third-party benchmark — while indexing 1% of the data.
Onton, a San Francisco-based search and discovery company, has released Ontology 1, a neurosymbolic model for complex, conversational, multimodal product search. On a 90-query benchmark scored by three independent LLM judges, Ontology 1 reached a mean precision@10 of 0.630, against 0.543 for Google …