The Nova 2.0 Omni (medium) is a unified multimodal reasoning model developed by Amazon and introduced in December 2025. It is part of the Amazon Nova 2 family, designed to handle text, images, video, and audio as inputs while natively generating both text and image outputs. The model uses a single, unified architecture to perform these tasks, reducing the complexity typically associated with coordinating multiple specialized models for multimodal workflows.
The "medium" designation refers to the model's hybrid reasoning configuration. This feature allows developers to adjust the depth of systematic, multi-step reasoning to optimize for specific latency, cost, and accuracy requirements. The medium setting is positioned as a balanced option for complex analytical and generative tasks that require sophisticated understanding without the higher operational costs of the maximum reasoning depth.
Key capabilities of the model include a 1-million-token context window, enabling the analysis of massive documents, long-form video, and large codebases in a single prompt. It supports text processing in over 200 languages and speech input in 10 languages. Its generative features include high-quality image creation and editing with natural language, supporting advanced functions like character consistency and text rendering within visual outputs.
Designed for enterprise-scale applications, the model excels in areas such as marketing content creation, speech transcription, and complex video metadata generation. It is a proprietary model offered through Amazon Bedrock, emphasizing safety, reliability, and customizability for business use cases.