New model outperforms OpenAI's CLIP and Google's SigLIP2, delivering measurably better shopping experiences and stronger revenue outcomes for publishers

SAN FRANCISCO, CA, UNITED STATES, June 2, 2026 /EINPresswire.com/ -- Shopsense AI announced today the Shoppable Intelligence Model (SIM), its next-generation proprietary multimodal AI powering real-time contextual commerce. Tested against the leading publicly available models, including OpenAI's CLIP and Google's SigLIP2, SIM delivers 25 to 50 percent higher retrieval accuracy across the benchmarks that matter most for product discovery. For publishers and their viewers, this model improvement delivers more relevant recommendations, higher audience engagement, and stronger commerce revenue, with no changes to existing integrations.

SIM is a proprietary in-house model built for commerce where every training decision, every dataset, and every benchmark is grounded in shopping outcomes: real products and real purchase behavior from where shopping decisions are formed - content. That specificity is what enables the model to differentiate itself from the more general-purpose industry benchmark models. On Fashion200K and FashionGen, two public fashion datasets used by the research community to track progress in retrieval AI, SIM outperforms every open-source baseline across every retrieval modality: image-to-image, image-to-text, text-to-image, and text-to-text. These are reproducible results against public benchmarks: Against OpenAI's CLIP, SIM achieves up to 77 percent higher image-to-image retrieval accuracy and up to 74 percent higher text-to-text accuracy on fashion datasets. Against Google's SigLIP2, SIM leads across retrieval tasks by 34 to 60 percent.

"The industry has been measuring progress in AI by how well models recognize content,” said People Inc. Chief eCommerce Officer Tory Brangham. "The real benchmark is whether they can monetize it. SIM raises that standard by tying intelligence directly to commerce outcomes, such as intent and conversion.”

What Better Retrieval Means in Practice
Product recommendations live or die on subtle visual cues (the cut of a sleeve, the texture of a fabric, the curve of a heel) and on the precise vocabulary used to describe them. A retrieval accuracy improvement of ten percentage points on a benchmark translates directly into recommendation quality improvements that drive shopper behavior changes. For every 100 customer searches, 10 additional shoppers see exactly what they were looking for as the first result. At the scale of a live publisher or retailer network, that compounds across click-through, conversion, and time-to-purchase. In Shopsense engagement data, this translates directly to a 10 percent improvement in model precision, which produces a 24.5 percent improvement in shopper click-through rate for native retailer media activations. For customers using SIM, that lift arrives automatically across every storefront and in-content commerce unit to deliver real revenue lift, with no additional effort.

SIM has been optimized across:
* Fashion and apparel: Wardrobe identification, style matching, and outfit-level product curation from editorial, broadcast, and social content
* Accessories: Bag, jewelry, eyewear, and footwear identification from images and video
* Furniture: Room-level product identification and style curation from editorial, images, and video
* Core Commerce: Cross-category product matching from editorial, product pages, and lifestyle content

"The gap between a general-purpose model and one that truly understands commerce is not incremental. It is the difference between surfacing a product a viewer might glance at and surfacing the product that inspires action," said Bryan Quinn, President and Co-founder at Shopsense AI. "Trained relentlessly for commerce. Built to understand intent. That's SIM."

This news appeared first in MediaPost:
https://www.mediapost.com/publications/article/415424/shoppable-ai-data-sets-new-standard-for-media-buye.html

About Shopsense AI
Shopsense AI turns content into commerce. Using a patent-pending agentic AI system, Shopsense transforms articles, videos, and broadcasts into seamless, personalized shopping moments. Publishers unlock premium revenue. Retailers extend their retail media footprint into moments of real consumer inspiration. Advertisers gain incremental performance at scale. Shopsense connects a catalog of 100 million daily products across 500,000 brands and 1,000+ retailers, and is live with Bell Media and CTV and across web publisher and creator networks.
Patent-pending systems: Feed Enrichment System (#63/564,250) and AI Recommendation System for Shoppable Experiences (#63/653,081).

Bill Brazell
Sharp Pen Media
+1 9174457316
email us here

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.