Amazon presents Nova, its new generation of multimodal artificial intelligence models

The company has developed six Nova models with unique capabilities: from text processing to image and video generation. Models like Nova Micro, Nova Lite, and Nova Pro stand out, offering cutting-edge performance with costs up to 75% more economical than their competitors on Amazon Bedrock.

The Nova models are notable for their versatility: they process text, image, and video, enabling the understanding of complex documents, multimedia content generation, and advanced reasoning tasks. The platform supports over 200 languages and offers customization capabilities through model fine-tuning and distillation.

Integration with Amazon Bedrock allows clients to experiment and evaluate different artificial intelligence models through a single API, facilitating adaptation to specific organizational systems and data.

Particularly innovative are Nova Canvas for image generation and Nova Reel for video creation, which are already being used by advertising teams to create more creative campaigns, multiplying the number of advertised products by five.

Amazon plans to expand its capabilities with two additional models in 2025: a speech-to-speech model that interprets verbal and non-verbal signals, and another multimodal "any-to-any" model capable of processing and generating content in multiple formats.