Pinterest cuts AI costs 90% by mixing open-source and proprietary models
Pinterest is shrinking its AI budget while expanding AI features, using a strategy that combines proprietary models it built in-house with free open-source software and paid closed-source systems from companies like OpenAI and Anthropic.
The approach has reduced costs by an estimated 90% compared to relying solely on Pinterest's own models, the company said during its February earnings presentation.
Vicky Gkiza, Pinterest's vice president of product management, said the "model-agnostic" strategy began in 2023. The company now uses OpenAI's language models for some product features, Anthropic's Claude for internal coding tasks, and Alibaba's open-source Qwen model for visual understanding and content labeling.
Why the mix matters
Closed-source models like OpenAI's process data quickly and integrate easily into existing systems, but they cost more per query. Open-source models are free to download and can be customized deeply, but they require engineers to build, maintain, and debug them.
Lan Guan, chief AI and data officer at Accenture, told Business Insider that companies are increasingly adopting this blended approach to manage token costs-the units of text that AI models process. "Open-source will be a really good option," Guan said.
New features in production
Pinterest rolled out two major features in 2025 using this mixed-model approach. Auto-collages lets advertisers convert product catalogs into pins for shopping feeds. The company tested the feature starting in early 2024 and launched a pilot with retailers like Macy's by June 2025.
A voice-enabled search feature, which uses both open-source and third-party AI, entered beta testing in October 2025. Early results showed users asked more shopping-specific questions when speaking rather than typing.
Building the team
To execute this strategy, Pinterest hired engineers with deep AI expertise. Matthias Zenger, a former Google engineer, joined as vice president of engineering in April 2025. Software engineer Mirjam Wattenhofer joined in August to focus on e-commerce and user experiences, and the company opened an Engineering Excellence Hub in Zurich.
Gkiza said the company is "investing more in hiring the right talent - evolving the team, whether it is engineering or product management - to be much more familiar with AI."
Infrastructure spending
To support open-source models, which require more processing power than managed services, Pinterest plans to invest in cloud infrastructure like graphics processing units.
The company said it will continue experimenting with different models: proprietary ones for personalization, open-source models for cost savings and visual tasks, and closed-source models when they deliver the best performance for specific use cases.
Learn more about Generative AI and LLM strategies and how they apply to AI for Product Development.
Your membership also unlocks:
