Shopify and Liquid AI announced a partnership to license and deploy the startup's flagship Liquid Foundation Models across Shopify workflows including search and other multimodal use cases.
Liquid AI is a Boston-area startup spun out of Massachusetts Institute of Technology that builds “liquid neural networks,” which are a form of foundation model designed to run efficiently on devices and embedded hardware. Shopify actually participated in Liquid AI's $250M Series A round in December 2024 and subsequently began co-developing its technology together for Shopify-specific use cases.
Liquid AI's models are “designed for sub-20 millisecond, multimodal, quality-preserving interference.”
In other words… Liquid AI's models are designed to give answers almost instantly, while still keeping the quality high, even when handling different types of input like text, images, or other data.
Shopify CTO Mikhail Parakhin said:
“I’ve seen a lot of models. No one else is delivering sub‑20ms inference on real workloads like this. Liquid’s architecture is efficient without sacrificing quality; in some use cases, a model with ~50% fewer parameters beats Alibaba Qwen, and Google Gemma, and still runs 2–10× faster. That’s what it takes to power interactive commerce at scale.”
As part of the partnership, Shopify and Liquid AI built a new recommendation engine together that beats its older recommendation tools and led to more shoppers clicking and buying the products it suggested, as well as a super-fast text model that helps improve search result speed and accuracy.
Shopify is also testing other Liquid AI models that can work with multiple types of data, such as customer profiles, product images, and store information, to power things like smarter recommendations, better product tagging, and more accurate customer insights.
Under the new agreement, Shopify will license Liquid AI’s foundation models, but the companies did not disclose the financial terms of the deal.