Nova Pro (EU)

Nova Pro sits in the middle of Amazon's Nova family as a balanced multimodal foundation model, positioned above the Lite tier for more demanding vision-and-language workloads. According to a third-party catalog entry, it processes text, image, and video inputs within the cataloged API limit context window, which makes it well suited to complex multi-turn interactions, retrieval-augmented generation workflows, and agentic applications that need to reason over long, mixed media. Its creator-targeted design emphasizes high accuracy on vision tasks while keeping language understanding and generation at practical speed and cost, distinguishing it from the longer-context "extended thinking" Nova 2 generation that Amazon later introduced.

Beyond raw multimodal ingestion, Nova Pro is reported to support document understanding with field extraction and data localization, visual inspection and defect detection, and video analysis with temporal reasoning, giving it a broad toolkit for enterprise-style automation. It also offers reliable function calling for tool use, prompt caching for repeated prompts, and fine-tuning so teams can adapt vocabulary and behavior to industry-specific terminology. The model family handles more than 200 languages with optimization concentrated on a core set including English, German, Spanish, French, Italian, Japanese, Korean, Arabic, Simplified Chinese, Russian, Hindi, Portuguese, Dutch, Turkish, and Hebrew, making it a practical choice for globally distributed products that mix vision, document, and conversational interfaces.