Pinterest is increasingly relying on Chinese open-source artificial intelligence models to improve its recommendation and shopping features, reflecting a broader shift among U.S. companies toward lower-cost, customizable AI tools developed outside the United States.
Pinterest Experiments With Chinese AI Models
Every month, hundreds of millions of users visit Pinterest to browse ideas and trends. The company has been testing Chinese AI models to refine how content is surfaced, even though it could rely on U.S.-based AI providers.
Chief Executive Officer Bill Ready said Pinterest has positioned itself as an AI-powered shopping assistant. He said the company’s increased use of Chinese models followed the release of DeepSeek R-1 in January 2025, which he described as a turning point because the model was open sourced.
Chinese AI developers whose models are being considered across the industry include Alibaba with its Qwen models, Moonshot with Kimi, and ByteDance, which is developing similar systems.
Cost And Customization Drive Adoption
Pinterest Chief Technology Officer Matt Madrigal said the appeal of these models lies in their open-source nature. Unlike many proprietary offerings from U.S. companies such as OpenAI, the models can be downloaded and tailored to Pinterest’s needs.
Madrigal said open-source techniques used to train Pinterest’s in-house models are about 30 percent more accurate than leading off-the-shelf alternatives. He also said the costs can be significantly lower, sometimes up to 90 percent less than proprietary models.
Broader Use Across U.S. Companies
Pinterest is not alone in adopting Chinese AI technology. These models are gaining traction among Fortune 500 companies.
Airbnb Chief Executive Officer Brian Chesky said in an October interview with Bloomberg that Airbnb relies heavily on Alibaba’s Qwen model for its AI customer service agent. Chesky cited its performance, speed, and cost.
Airbnb has said it uses multiple AI models, including U.S.-based ones, and hosts them within its own infrastructure so that user data is not shared with model developers.
Open-Source Momentum On Hugging Face
The growing popularity of Chinese models is also visible on Hugging Face, where developers download and share AI systems. Jeff Boudier, who builds products at Hugging Face, said cost is a major factor for startups choosing Chinese models.
Boudier said Chinese models often occupy many of the top positions among the most downloaded and liked training models. He noted that there are weeks when four of the top five models come from Chinese labs.
In September, Alibaba’s Qwen surpassed Meta’s Llama family to become the most downloaded large language model group on the platform.
Shifts Among U.S. AI Developers
Meta released its open-source Llama models in 2023, which were widely used by developers before the arrival of DeepSeek and Alibaba’s offerings. Developers later expressed disappointment with Llama 4, released last year.
Meta has since been reported to be using open-source models from Alibaba, Google, and OpenAI to help train a new model set planned for release this spring.
Changing Perceptions Of AI Leadership
A report published last month by Stanford University said Chinese AI models appear to have matched or exceeded global peers in capability and adoption. The report also suggested that government support may have contributed to China’s progress in open-source AI.
In a recent interview with the BBC, former UK deputy prime minister Nick Clegg said U.S. firms have been focused on developing AI that could surpass human intelligence, while China has emphasized more accessible technologies.
Clegg, who left his role as head of global affairs at Meta last year, said China is doing more to broaden access to the technology it is competing over.
Revenue Pressures On U.S. AI Firms
U.S. AI companies are facing pressure to generate revenue as they invest heavily in computing infrastructure. OpenAI released two open-source models last summer but has focused most of its resources on proprietary systems.
OpenAI Chief Executive Officer Sam Altman said in October that the company is investing aggressively in training and infrastructure, adding that revenue is expected to grow quickly alongside those investments.
Featured image credits: Plann
For more stories like it, click the +Follow button at the top of this page to follow us.