Nova Pro

Amazon's Nova Pro v1 sits in the middle of Amazon's Nova model family, positioned between the budget-oriented Nova Lite and Nova Micro variants and the top-tier Nova Premier. Independent reporting frames it as a proprietary, closed-weight system designed to balance capability with cost, costing notably less than frontier-tier models while keeping latency low for production use. This positioning makes it well suited to applications where dozens of inference calls happen per task, such as agentic workflows and retrieval-augmented generation pipelines, rather than workloads that demand the absolute highest reasoning accuracy.

Nova Pro v1 is a multimodal model handling both text and image inputs, and independent evaluators have highlighted its reliability for function calling, long-range planning, and competitive scores on Berkeley-style tool-use benchmarks. In head-to-head RAG testing it answered questions faster and more cheaply than competing systems at comparable accuracy, and it handled summarization tasks with substantial per-request savings while preserving output quality. Taken together, these characteristics describe a mid-tier model aimed at teams who want dependable multimodal understanding and agentic behavior without paying frontier-model prices.