Nova 2Intelligent document assistant
Process large documents with up to 1 million tokens of context to answer questions and extract insights (with RAG).
Amazon Nova 2
Nova 2
Multimodal foundation models for text, images, video, documents and speech
Amazon Nova provides multimodal foundation models that process text, images, video, documents and speech. With support for up to 1 million tokens of context and advanced reasoning capabilities, Amazon Nova 2 models enable you to build sophisticated AI applications that understand complex inputs and generate accurate responses.
Multimodal1M contextAdvanced reasoningSpeech & textFlexible deployment
Create with Nova 2
Chat, analyze, reason and speak — in a single multimodal experience
You can build interactive chatbots, analyze documents and videos, create AI agents with extended reasoning and develop voice-enabled applications.
- Context windowUp to 1 million tokens
- Maximum output65,536 tokens per response
- ReasoningExtended thinking with step-by-step analysis
- DeploymentAccessed through Amazon Bedrock
TextImagesVideoDocumentsSpeech
Sign in with Google to generate. Accounts with credits continue to the Nova 2 chat experience; plans and credits are shown on the pricing page.
Three models, each optimized for different use cases
Amazon Nova 2 includes the following models, each optimized for different use cases.
ModelInput modalitiesOutput modalitiesUse cases
Nova 2 LiteText, images, video, documentsTextHigh-volume applications prioritizing speed and cost efficiency
Nova 2 SonicSpeech, textSpeech, textVoice-enabled applications with fast response times
Nova Multimodal EmbeddingsText, images, documents, video, audioEmbeddingsSemantic search, recommendation systems and similarity matching
All models support up to 1 million tokens of context and can generate up to 65,536 tokens in a single response. Models with reasoning capabilities can perform extended thinking to solve complex problems step by step.
Sophisticated applications on Amazon Nova
The following are examples of what you can build with Amazon Nova.
Complex reasoning applications
Solve multi-step problems with extended thinking that shows the model’s step-by-step analysis (or with reasoning).
Video analysis pipeline
Extract insights, generate summaries and identify key moments in video content at scale.
Voice-enabled AI agent
Build conversational agents that understand speech input and respond with natural language.
Why teams choose Amazon Nova
Amazon Nova provides the following benefits.
Multimodal understanding
Process text, images, video, documents and speech in a single request. Amazon Nova models understand relationships across different input types.
Extended context
Support for up to 1 million tokens allows you to process entire codebases, lengthy documents, or extended conversations without losing context.
Advanced reasoning
Models with reasoning capabilities break down complex problems and show step-by-step analysis, improving accuracy for multi-step tasks.
Flexible deployment
Access models through Amazon Bedrock with no infrastructure to manage, or customize models through fine-tuning and reinforcement learning.
Built-in tools
Use web grounding to access real-time information and code interpreter to execute Python code without external integrations.
The basic workflow
Amazon Nova models are foundation models that you access through Amazon Bedrock.
Step 1
Your application sends a request to Amazon Bedrock with your input and configuration parameters.
Step 2
The Amazon Nova model processes your input, applying reasoning if configured.
Step 3
The model generates a response and returns it to your application.
Step 4
You can enhance responses by using RAG to incorporate your data, enabling built-in tools, or customizing models through fine-tuning.
Priced on tokens, optimized per model
Amazon Nova pricing is based on input and output tokens processed. Different models have different pricing tiers.
Nova 2 Lite
Optimized for cost-effective, high-volume processing
Text, images, video and documents in; text out. Built for speed and cost efficiency at scale.
Nova 2 Sonic
Balanced pricing for voice-enabled applications
Speech and text in; speech and text out. Built for voice agents with fast response times.
For current pricing information, see Amazon Bedrock Pricing. Compare chat plans and credit packs on the Nova 2 pricing page.
Nova 2 key concepts
Before you learn about Amazon Nova models, familiarize yourself with the following core concepts.
Foundation models+
Pre-trained AI systems available in different sizes and capabilities that you access through an API.
Inference+
The process of sending a request to a model and receiving a generated response.
Reasoning+
Extended thinking capability that allows models to break down complex problems and show their step-by-step analysis before providing answers.
Multimodal+
The ability to process and understand multiple input types together: text, images, video and documents in a single request.
RAG (Retrieval-Augmented Generation)+
A technique that combines model responses with your own data sources to provide more accurate, contextual answers.
Start creating with Nova 2
Sign in with Google and continue into the Nova 2 chat experience. Interactive chatbots, document and video analysis, agents with extended reasoning, and voice-enabled applications — all in one conversation.