Gemma 3 4B IT (Amazon Bedrock, US)

Documentation evidence for this exact Gemma 3 4B IT variant on Amazon Bedrock is currently limited: the supplied source consists of cookie-preference page text rather than substantive technical content. Catalog identity fields place the model within Google's Gemma family as a multimodal, open-weights entry intended for instruction-following tasks across text and image inputs with text outputs. Because no architecture, parameter count, training data, or benchmark results for this specific 4B IT variant are corroborated by the supplied excerpts, those specifics are omitted here rather than restated from unverified memory.

For practitioners, the practical framing is therefore necessarily narrow: the model is presented as a smaller, open-weights member of the Gemma 3 lineup suitable for cost-sensitive or on-device style deployments where instruction tuning and multimodal input are useful, while detailed evaluation against peers or guidance on ideal prompting patterns would require additional creator-side documentation. Users evaluating this Bedrock-hosted variant should treat current public documentation as a thin layer over the model and seek Google's Gemma 3 technical material for fuller context on capabilities and intended use.