Amazon Nova 2: Multimodal Reasoning and Generation Models

Xiaohan Fei

Abstract:

We present Amazon Nova 2, a family of four foundation models designed to meet diverse enterprise
needs across reasoning, multimodal processing, and real-time conversational AI. The family includes
Nova 2 Lite and Nova 2 Pro — multimodal models with dynamic reasoning capabilities that allow
customers to balance accuracy, speed, and efficiency through configurable “extended thinking”
controls; Nova 2 Omni — a unified multimodal model that processes text, images, video, and audio
inputs while generating both text and images; and Nova 2 Sonic — a speech-to-speech foundation
model for natural conversational AI. Nova 2 models process large contexts with up to 1M tokens,
enabling analysis of extensive codebases, long documents, and videos within a single prompt. Like
all Nova models, the Nova 2 family is built with integrated safety measures and Responsible AI
guardrails, maintaining our commitment to customer trust, security, and reliability