Inside Pinterest's efforts to replace expensive AI with open-source models (www.businessinsider.com)

Pinterest is shifting its AI strategy by blending open-source and closed-source models, leading to significant cost reductions while enhancing its platform's functionality. In a recent update, the company reported that its mixed approach—leveraging proprietary technology alongside models from OpenAI, Anthropic, and Alibaba—has resulted in AI-enabled features that are approximately 90% cheaper to operate compared to relying solely on proprietary models. This strategy is part of a broader trend where companies in the AI/ML community are pressured to demonstrate value from their investments by optimizing costs associated with AI systems. Key elements of Pinterest's strategy include investing in cloud infrastructure and talent acquisition focused on AI and machine learning expertise. By mid-2025, the platform plans to roll out innovative features like auto-collages for product catalogs and voice-enabled AI capabilities that respond to user queries, showcasing the effectiveness of its hybrid AI model. This approach not only enhances user engagement but also reduces token costs, making it a compelling model for the industry as organizations seek to balance performance with financial sustainability.