AWS adds NVIDIA Nemotron 3.5 Super VL vision-language model to SageMaker JumpStart

AWS announced that NVIDIA's Nemotron 3.5 Super VL is available through Amazon SageMaker JumpStart, deployable from SageMaker Studio or with the SageMaker Python SDK. The vision-language model takes images, video and text as input and returns text, pairing a 120B-parameter hybrid mixture-of-experts backbone with 12B active parameters and a C-RADIO vision encoder; context can reach 1 million tokens.

JumpStart lists two checkpoints: an NVFP4 build quantized from BF16 for optimized inference on G7 instances, and full BF16 reference weights on P5en for teams that want to fine-tune, distill or build their own quantized variants. AWS describes the model as aimed at document intelligence, screen grounding for computer-use agents, video question answering and multimodal agents, with tool calling, JSON schema output and configurable reasoning. It ships under the OpenMDW 1.1 license.