Amazon Nova foundation model
Amazon Nova Micro
Text-only Amazon Nova understanding model for the lowest-cost, lowest-latency text tasks through Amazon Bedrock.
Key points
- Accepts text input and outputs text; it is not multimodal.
- Is the cheapest, fastest, and lowest-latency Nova understanding model.
- Supports about a 128K token context window.
- Supports more than 200 languages, with optimization for a core language set.
- Supports fine-tuning for supported customization use cases.
When to use it
- Choose Micro for high-volume classification, routing, simple Q&A, and summarization.
- Use it when latency and cost matter more than multimodal input or complex reasoning.
- Use through Amazon Bedrock when a text-only foundation model is sufficient.
Exam tips
- Micro is the only Nova understanding model that is text-only.
- Micro is the lowest-cost and fastest Nova understanding model.
- Do not choose Micro when the prompt must include images, video, or speech.