Getting Started

eIQ® is a comprehensive development environment for artificial intelligence, designed to support both simple, targeted tasks and end‑to‑end, advanced workflows. It comprises a modular set of tools that can be installed and used independently via command-line or Python interfaces, enabling lightweight, focused operations.

For more advanced use cases, eIQ seamlessly combines these components into higher-level solutions such as eIQ Olive (available via CLI and Python) and the eIQ AI Toolkit (offering graphical and REST interfaces). These tools provide advanced analytics, optimization, and profiling capabilities. To further lower the entry barrier, eIQ also offers eIQ AI Hub, a web-based cloud service.

The following diagram shows individual components documented in eIQ Training Materials.

eIQ Overview

Tools Overview

eIQ AI Hub

eIQ AI Hub is NXP’s cloud-based platform that unifies the entire edge-AI development lifecycle into one integrated environment. It provides instant access to the latest eIQ tools, cloud-connected hardware, hosted services, and curated resources—removing the traditional complexity of setting up toolchains, preparing hardware, or managing dependencies.

Built for both rapid experimentation and production-grade development, eIQ AI Hub enables developers to transform ideas into deployable AI solutions faster, more reliably, and with significantly less overhead.

eIQ AI Toolkit

eIQ AI Toolkit is an advanced tool offering graphical and REST API interfaces for two primary purposes:

  • Optimizing, converting and deploying neural network models through structured workflows

  • Profiling neural network models

These capabilities enable users to fine-tune their models using a wide selection of algorithms, identify and resolve deployment issues, validate results, and measure model performance.

Note

See the eIQ AI Toolkit Video Tutorials on how-to start with the GUI.

eIQ Neutron SDK

eIQ Neutron SDK provides tools, drivers, and firmware for enabling or developing with the eIQ Neutron NPU. It is available in two formats:

  • A complete archive containing all assets—from tools to runtime software—for MCU and MPU enablement

  • A Python package for integration into third-party applications

The runtime includes MPU drivers, MCU/MPU firmware, and a TensorFlow Lite delegate. Additionally, the SDK offers three key tools:

  • Converter – Compiles models into specialized TF Lite files embedding Neutron microcode for deployment

  • Quantizer – Performs quantization of TF Lite models in Neutron-specific formats

  • Profiler – Provides detailed insights into model deployment on Neutron

eIQ Olive

eIQ Olive is an open-source optimization framework extended for NXP hardware. It incorporates the optimization and deployment capabilities of eIQ AI Toolkit, offering full functionality through CLI and Python interfaces. However, unlike eIQ AI Studio, it does not include integrated checks or additional features such as profiling.

eIQ ONNX2TFLite

A Python module that converts ONNX models directly to TF Lite without intermediate formats. It includes **eIQ ONNX2Quant ** for quantization in ONNX format, ensuring correct settings for eIQ Neutron NPU. Together, these tools can apply additional optimizations during conversion.

eIQ Vela for i.MX 93

A compiler for the Arm Ethos-U accelerator available on NXP i.MX 93 microprocessors.