Nova Micro 1.0

Nova Micro 1 sits within the Amazon Nova family as a text-only chat model built for speed and economy, offered through Amazon Bedrock with a 128,000-token context window and native tool use support. The design centers on latency-sensitive, short-form workloads such as classification, retrieval, and lightweight agentic tool-calling loops rather than long-form generation or multimodal reasoning. According to the available evidence, generation speeds can exceed 200 tokens per second, and the model is positioned as the fastest, lowest-cost member of the Nova lineup, making it well suited to high-volume production traffic where responsiveness matters more than depth. The practical ceiling for single responses is 5,120 tokens, which routes longer synthesis tasks to larger siblings like Nova Pro or Nova Premier in a multi-model pipeline.

For builders, Nova Micro 1 is best slotted into the front end of a pipeline, where it can handle real-time translation, document classification over long inputs, and fast extraction or routing decisions before passing work to a heavier model. Its pricing and throughput profile support cost-sensitive deployments such as CRM enrichment, outreach triage, and interactive chat routing. The model is text-only on input, so tasks requiring image understanding or extended generation are out of scope, and a knowledge cutoff of October 2024 is reported by the source. Native Bedrock integration means teams already using AWS can wire it in with existing IAM, logging, and governance tooling.