Related Software

NVIDIA NIM for Large Language Models (LLM) and Vision Language Models (VLM) fits into a broader inference and platform ecosystem. The following software products are highly relevant when you are deploying, operating, or extending LLM workloads. For capabilities of NIM LLM and VLM itself, refer to Overview.

NVIDIA NIM Within the Inference Ecosystem

The container image you deploy determines which inference engine NIM LLM and VLM runs. The following table lists related products:

Product	Relation to NIM LLM and VLM
vLLM	The inference engine in the vLLM-based NVIDIA NIM for Large Language Models (LLM) and Vision Language Models (VLM) container images. Request semantics and tuning concepts for those images come from vLLM.
SGLang	The inference engine in the SGLang-based NVIDIA NIM for Large Language Models (LLM) and Vision Language Models (VLM) container images. Request semantics and tuning concepts for those images come from SGLang.
NIM Operator	The operator manages NIM deployments by using Kubernetes custom resources and is especially useful for repeatable, production-scale rollouts.
NVIDIA NIM for Vision Language Models	Earlier VLM-specific documentation remains useful for deployments that still follow that path. Use these NIM LLM and VLM docs for the combined LLM and VLM workflow.

Choosing Related Software

To help you choose the right tool for your specific use case, consider the following recommendations:

• Use NIM LLM and VLM when your workload uses text, image, audio, or video input and you want a curated, enterprise-ready container for production inference. Modality support depends on the inference backend in your image. Refer to Image, Audio, and Video Input.

• Use the documentation for the inference engine in your container image (vLLM or SGLang) alongside NIM documentation when you need deeper, backend-specific context for passthrough arguments or upstream model-serving behavior. Passthrough arguments must be valid for the engine in the image you run. Refer to Advanced Configuration for the per-backend flag tables.

• Use the NIM Operator when your primary deployment target is Kubernetes and you want lifecycle automation around NIM services.