NVIDIA Nemotron 3.5 Lightning model is now available on Amazon SageMaker JumpStart

NVIDIA has made its Nemotron 3.5 Lightning model available on Amazon SageMaker JumpStart, enabling AWS customers to deploy it for persistent agent workloads. The model is designed for high-throughput automation across sectors such as personal assistants, financial document processing, and cybersecurity triage. It operates on a hybrid Mixture-of-Experts architecture with 30 billion total parameters, activating only 3 billion per forward pass to optimize performance. The model achieves up to four times the throughput and 30% faster task completion compared to comparable models in its class.

The model supports up to 1 million tokens of context through DFlash speculative decoding and integrates with popular agent frameworks. It is fully open-trained on open datasets, allowing enterprises to customize it for their specific tools, workflows, and policies. This flexibility enables deployment across edge, on-premises, or cloud environments with full ownership retained by the user. Nemotron 3.5 Lightning is positioned as a cost-effective solution for enterprises requiring scalable AI automation.

AWS customers can deploy the model through SageMaker JumpStart with minimal setup, either via the SageMaker console or the SageMaker Python SDK. The deployment process is streamlined to reduce operational overhead, making it accessible for organizations without extensive AI infrastructure. This integration aligns with AWS’s strategy to provide ready-to-use AI solutions for enterprise workloads.

The availability of Nemotron 3.5 Lightning on SageMaker JumpStart follows NVIDIA’s broader push to expand access to high-performance open models. Enterprises can now leverage this model for rapid task execution and persistent agent applications without proprietary licensing constraints. For detailed deployment instructions, users are directed to the Amazon SageMaker JumpStart documentation.