.svg){kind=spacer}
Details
- NVIDIA has announced the release of its open-weight Nemotron Nano 2 and multimodal Nano 2 VL foundation models as fully managed endpoints on Amazon Bedrock.
- These models allow developers to create text, code, images, and even short videos through a single API, supporting advanced use cases like autonomous customer-support bots and automated security workflows.
- Nemotron Nano 2 models are designed for efficiency, boasting a small footprint with fewer than 5 billion parameters and optimized for rapid inference on NVIDIA GPUs; the VL version adds advanced vision-language capabilities.
- Amazon Bedrock users now benefit from automatic scaling, strong encryption, and pay-as-you-go pricing, with these models joining offerings from Anthropic, Meta, Mistral, and Amazon’s Titan series.
- Early adopters such as CrowdStrike and BridgeWiseAI are leveraging the tech for cybersecurity telemetry enrichment and equity research summarization, respectively.
- The Nemotron Nano 2 models are immediately available in US-East and EU-West regions, with plans for a global rollout in the coming weeks as demonstrated during the launch event.
Impact
This expansion enhances Amazon Bedrock’s selection, heightening its rivalry with Microsoft Azure’s OpenAI Service and Google Vertex AI, especially as enterprises seek multimodal AI options. The lightweight, cost-efficient design could make such models more attractive for organizations aiming to lower operational costs. Closer collaboration between NVIDIA and Amazon spotlights their growing AI partnership, while open-weight availability aligns with evolving regulatory and transparency demands in the industry.
