Synopsis: In this Techstrong AI video interview, Vasi Philomin, vice president of generative artificial intelligence (AI) for Amazon Web Services (AWS), explains why the cloud service provider is providing access to a less expensive family of Nova large language models (LLMs).
Amazon Web Services (AWS) has launched its new Nova large language models (LLMs) to address the evolving needs of businesses leveraging AI. Vasi Philomin, VP of generative AI at AWS, explains that Amazon’s diverse business operations provided insight into gaps in existing AI models, leading to the development of Nova. These models are designed to be both high-performing and cost-efficient, making them suitable for large-scale enterprise applications. With an industry-leading price-performance ratio, the Nova models aim to offer businesses an affordable yet powerful alternative in the AI space.
AWS’s Nova portfolio includes multiple models tailored for different applications. The four primary text-based models—Nova Micro, Nova Lite, Nova Pro, and Nova Premier—support text inputs and, in most cases, images and video. Additionally, AWS has introduced Nova Canvas for image generation and Nova Real for video creation, with upcoming models set to support speech-to-speech interactions and fully multimodal capabilities. These models offer advanced functionality while maintaining affordability, making them particularly appealing for enterprises looking to scale AI adoption.
To optimize costs and efficiency, AWS is focusing on innovative AI deployment strategies, including model distillation and custom-built hardware like Trainium and Inferentia chips. Philomin highlights that businesses can dynamically route AI queries to the most appropriate model based on complexity, reducing computational overhead. As AI models continue to evolve in complexity and efficiency, AWS is committed to reducing costs and improving scalability, ensuring that enterprises can integrate AI seamlessly into their operations without excessive financial strain.
