Skip to main content
NexaSDK Docker Architecture

​
Linux Docker Solution

NexaSDK Docker enables you to run AI models in containerized environments on Linux ARM64 systems, optimized for Qualcomm NPU devices like Dragonwing IQ9. This solution provides a consistent, isolated environment for running inference workloads with support for both server mode (REST API) and interactive CLI usage.

​
Key Features

  • Containerized Deployment: Run NexaSDK in isolated Docker containers
  • NPU Acceleration: Optimized for Qualcomm NPU on Linux ARM64 (Dragonwing IQ9)
  • Dual Modes: Support for both server mode (REST API) and interactive CLI
  • Easy Setup: Simple pull and run commands
  • Multiple Model Types: Support for LLM, Embeddings, Reranking, Computer Vision, and ASR models

​
Supported Models

The following models are supported on Linux ARM64 with NPU acceleration (Dragonwing IQ9):
ModalityModel
LLMNexaAI/LFM2-1.2B-npu
NexaAI/Granite-4.0-h-350M-NPU
VLMNexaAI/OmniNeural-4B
EmbeddingNexaAI/embeddinggemma-300m-npu
NexaAI/EmbedNeural
RerankNexaAI/jina-v2-rerank-npu
CVNexaAI/yolov12-npu
NexaAI/convnext-tiny-npu-IoT
ASRNexaAI/parakeet-tdt-0.6b-v3-npu
For more information about NPU models and access tokens, see the NPU Models Guide.

​
Docker Hub

The NexaSDK Docker image is available on Docker Hub:

Docker Hub Repository

View available tags and pull the latest NexaSDK Docker image

​
Get Started

Quickstart for Docker

Get started with NexaSDK Docker in minutes. Learn how to pull the image, run in server mode, and use the CLI.