Qwen3.6-35B-A3B NIM Container for DGX Spark Overview

Description

The Qwen3.6-35B-A3B NIM Container for DGX Spark is a deployable inference container for serving Qwen3.6-35B-A3B, a third-party multimodal Mixture of Experts model capable of processing text, image, and video inputs for text generation. The container provides an OpenAI-compatible API for self-hosted deployment on NVIDIA DGX Spark, powered by the vLLM backend, and ships multiple precision and speculative-decoding presets selectable at run time.

The container components are ready for commercial use.

Third-Party Community Consideration

The model embedded in the container is not owned or developed by NVIDIA. This model has been developed and built to a third-party's requirements for this application and use case; see link to Non-NVIDIA Qwen3.6-35B-A3B Model Card.

License/Terms of Use:

GOVERNING TERMS: The NIM container is governed by the NVIDIA Software License Agreement and the Product-Specific Terms for NVIDIA AI Products; and the use of the model is governed by the NVIDIA Open Model License Agreement.

Additional Information: Apache-2.0 License.

You are responsible for ensuring that your use of NVIDIA provided models complies with all applicable laws.

Program Classes:

The Qwen3.6-35B-A3B NIM Container for DGX Spark includes the following models:

Model Name & Link	Use Case	How to Pull the Model
(Default) Qwen3.6-35B-A3B-FP8	FP8 target model for multimodal text generation, agentic coding, visual understanding, reasoning	Automatic (downloaded at startup with per-file checksum verification)
Qwen3.6-35B-A3B-DFlash	DFlash drafter model used for draft-model speculative decoding to accelerate the target model	Automatic (downloaded at startup with per-file checksum verification)
Qwen3.6-35B-A3B-NVFP4	NVFP4 target model for reduced-memory, higher-throughput multimodal inference	Automatic (downloaded at startup with per-file checksum verification)

Preset Configurations:

Each precision ships with two speculative-decoding presets, selected at run time via the NIM_MODEL_PROFILE environment variable:

MTP preset — uses the built-in Multi-Token Prediction (MTP) head of Qwen3.6-35B-A3B as the speculative drafter; no extra draft model is required. This is the default preset.
DFlash preset — uses the separate Qwen3.6-35B-A3B-DFlash drafter model for block-wise draft-model speculative decoding, providing the best single-stream generation latency.

Supported GPUs:

DGX Spark*1

Deployment Details

The Qwen3.6-35B-A3B NIM Container for DGX Spark exposes an OpenAI-compatible chat completions API for seamless integration into existing applications and workflows.

API Endpoints:

/v1/chat/completions — Chat completions (streaming and non-streaming)
/v1/models — List available models
/v1/health/ready — Health check

Operating System: Linux

Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.

Reference(s):

Container Version(s):

1.7.0-variant

Ethical Considerations:

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal developer team to ensure these software components meet requirements for the relevant industry and use case and address unforeseen product misuse.

Please make sure you have proper rights and permissions for all input image and video content; if image or video includes people, personal health information, or intellectual property, the image or video generated will not blur or maintain proportions of image subjects included.

Please report quality, risk, security vulnerabilities or NVIDIA AI Concerns here.

Get Help

Getting started with the NIM

Deploying and integrating the NIM is straightforward thanks to our industry standard APIs. Visit the NIM Overview page for release documentation, deployment guides and more.

NVIDIA Developer Community Forum

Get access to community knowledge base articles and support cases (https://forums.developer.nvidia.com/)

Qwen3.6-35B-A3B-Spark NIM Certified 容器镜像 公开资源