# nndeploy/nndeploy

一款简单易用和高性能的AI部署框架 | An Easy-to-Use and High-Performance AI Deployment Framework

Repository: https://github.com/nndeploy/nndeploy
Canonical: https://ross.abutalabs.com/products/nndeploy
Homepage: https://nndeploy-zh.readthedocs.io/zh-cn/latest/
Language: C++
License: Apache-2.0
License Family: permissive
Topics: ai, ascend, deep-learning, deployment, diffusers, genai, llm, low-code, low-code-platform, mnn, no-code, onnxruntime, openvino, python, pytorch, tensorrt, transformer, workflow
Last push: 2026-08-15T13:56:15+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 78, longevity 80
- inputs: {"age_days": 1121, "days_push": 18, "days_rel": 151, "gap_med": 2.0, "n_releases_24m": 13}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1868, forks 233 (observed 2026-08-28T04:05:46.415253+00:00)

## What it is
nndeploy is an easy-to-use, high-performance AI deployment framework written in C++ with Python bindings. It provides a visual drag-and-drop workflow for building AI pipelines that can be exported as JSON and deployed across desktop, mobile, edge devices, and servers using multiple inference backends like TensorRT, ONNX Runtime, OpenVINO, and MNN.

## Use cases
- deploy deep learning models on edge devices like Jetson and Ascend
- build AI pipelines with a drag-and-drop visual workflow
- run LLMs and diffusion models with a node-based interface
- export an AI workflow as JSON and call it from C++ or Python
- deploy the same model across Windows, Android, iOS, and Linux
- switch inference backends like TensorRT, OpenVINO, or ONNX Runtime without rewriting code
- write custom preprocessing nodes in Python or high-performance nodes in C++/CUDA

## When to choose
- you need to deploy AI models across heterogeneous hardware (desktop, mobile, edge, server)
- you want a visual, low-code workflow for composing AI pipelines
- you need multi-backend inference support without vendor lock-in
- you are deploying LLMs or AIGC models and want a workflow-based tool
- you want workflows exportable as JSON for production integration

## When to avoid
- you only need a single-backend inference runtime with a mature ecosystem like TensorRT alone
- you need a battle-tested framework with large community support like ONNX Runtime or TFLite
- your project is pure training rather than deployment
- you require extensive third-party integrations beyond what the framework currently offers

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, llm-inference, workflow-automation, deployment, gpu-computing, developer-tools
- domain: deep-learning, large-language-models, machine-learning, developer-tools, cross-platform
- platform: windows, python, cpp, cross-platform
- tags: ai-deployment, inference-engine, visual-workflow, low-code, tensorrt, onnxruntime, openvino, mnn, edge-deployment, stable-diffusion, linux, macos, android, ios, gpu

## Member repositories
- nndeploy/nndeploy (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:46.415253+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:15:16.089198+00:00, confidence not recorded.
  - readme: https://github.com/nndeploy/nndeploy (fetched 2026-08-28T04:05:46.415253+00:00, sha 6e3551a23ff8)
- Data as of 2026-08-30T08:39:29.467469+00:00.