Nova 2Amazon Nova 2 is a family of multimodal foundation models. Instead of one model for text and another for images or audio, Nova 2 understands several input types at once and answers in a single, coherent response.
What makes Nova 2 different
- Multimodal understanding — text, images, video, documents and speech can be sent in the same request.
- Up to 1 million tokens of context — long documents, whole codebases and extended conversations fit in one window.
- Extended reasoning — reasoning models break multi-step problems down and show their step-by-step analysis.
- Voice in and out — Nova 2 Sonic understands speech input and answers with natural language.
- No infrastructure to manage — models are reached through an API, with no servers to run.
The models
| Model | Input modalities | Output modalities | Best for |
|---|---|---|---|
| Nova 2 Lite | Text, images, video, documents | Text | High-volume applications that prioritize speed and cost efficiency |
| Nova 2 Sonic | Speech, text | Speech, text | Voice-enabled applications with fast response times |
| Nova Multimodal Embeddings | Text, images, documents, video, audio | Embeddings | Semantic search, recommendation systems and similarity matching |
What you can build
Intelligent document assistants. Feed long contracts, reports or manuals and ask questions across the whole document with RAG.
Complex reasoning applications. Let the model plan, check its own work and explain the steps it took.
Video analysis pipelines. Summarize long recordings and find the moments that matter at scale.
Voice-enabled agents. Build assistants that listen and reply in natural language.
Key concepts
Foundation models are pre-trained AI systems you access through an API. Inference is the act of sending a request and receiving a generated response. RAG (Retrieval-Augmented Generation) combines model answers with your own data sources to keep responses accurate and grounded.
Start creating with Nova 2
Sign in with Google and open the chat experience to try Nova 2 on your own documents, images and video. When you need more capacity, compare credit packs and plans.