At Amazon’s annual AWS re:Invent conference, the company announced Nova, a new line of generative AI models that aim to compete with OpenAI, Google, and Anthropic.

The suite, which can create text, images, and videos, as well as a speech-to-speech model that should be ready in 2025, was announced by CEO Andy Jassy. The models are available through Amazon Bedrock, the company’s managed service for AI integration.

Meet Nova

The Nova lineup includes:

• Nova Micro: Text-only model optimized for low-cost, fast responses.

• Nova Lite: A multimodal model for text, image, and video tasks at minimal cost.

• Nova Pro: Balances speed, cost, and accuracy across tasks.

• Nova Premier: Advanced reasoning and model distillation (launching Q1 2025).

• Nova Canvas: Creative image generation.

• Nova Reel: High-quality video generation, outperforming competitors in testing. 

With support for 200+ languages, Nova is optimized for RAG, fine-tuning, and knowledge distillation, simplifying adaptation of the tools for proprietary data by businesses.

The Nova models serve advertisers and developers, fueling innovations such as creative optimization and video advertising. Amazon emphasizes safety and transparency with AWS AI Service Cards and built-in features such as watermarking and content moderation.

Moreover, with Amazon expanding Nova with speech and multi-modality models, the entry of the company into generative AI is a bold step in the journey for scalable, real-world AI solutions.