Kneron

Developer Workbench

Kneron Toolchain v0.14.0 Released

Build AI on the Edge. Fast.

Convert, optimize, and deploy PyTorch/ONNX models to Kneron NPUs in minutes using our new unified Python SDK.

Start 5-Minute GuideBrowse Model Zoo
bash
# 1. Install the toolchain
$ pip install kneron-toolchain
# 2. Compile your model
$ kneron build --model yolo.onnx --target KL720
[INFO] Optimizing graph... [OK] [INFO] Quantizing to INT8... [OK] [SUCCESS] Generated 'yolo_kl720.nef'
# 3. Run inference
$ python app.py

NeuWare Studio

Code
K-Code Extension

Develop, debug, and flash firmware directly within VS Code. Integrates Docker seamlessly.

Profile
K-Scope Profiler

Visualize model latency layer-by-layer. Identify memory bottlenecks and power spikes.

Optimize
K-Surgeon

Visual graph editor for ONNX. Fix unsupported operators and preview quantization drift.

Common Tasks

Quick Start

Setup hardware & run Hello World.

Convert Model

PyTorch/TF -> ONNX -> NEF.

API Reference

Python & C++ SDK documentation.

Update Firmware

Flash the latest NPU/SCPU code.

Latest Resources
View All
KNEO Pi ACE User Guide v1.0.0

Released 2025-11-20 • Manuals

KL730 96Board Firmware SDK v1.3.0

Released 2025-07-18 • SDKs

Kneron Yolov7 Modification Guide v1.3

Released 2025-06-09 • Tutorials

Join the Community

Connect with other developers, share your projects, and get help directly from Kneron engineers.

Join Discord Server
Also Find Us On