Qualcomm AI Hub

The platform for on-device AI

Contact for Pricing

Description

Qualcomm AI Hub provides a comprehensive platform designed to streamline the deployment of artificial intelligence models directly onto edge devices. It facilitates bringing your own models or utilizing a library of pre-optimized models from various partners. The platform supports developers throughout the AI workflow, from model discovery and optimization to on-device deployment and performance analysis.

Leveraging the Qualcomm AI Stack, the Hub allows users to compile and optimize models for specific Qualcomm hardware, targeting the CPU, GPU, or NPU for optimal performance using runtimes like TensorFlow Lite and ONNX Runtime. It integrates with an ecosystem of model makers, cloud providers, and ML service partners to create end-to-end solutions for industries including mobile, compute, automotive, and IoT, enabling the development of next-generation AI-powered user experiences.

Key Features

Model Hub: Access pre-optimized models from partners like Mistral AI, G42, and Llama3.2.
Bring Your Own Model (BYOM): Upload and optimize custom AI models.
Model Compilation: Convert and optimize models for specific Qualcomm hardware and runtimes (TensorFlow Lite, ONNX Runtime, AI Engine Direct).
On-Device Profiling: Analyze model performance including compute unit usage, latency, and memory metrics on physical hardware.
Cross-Device Deployment: Deploy models to mobile, compute, automotive, and IoT devices powered by Qualcomm.
Ecosystem Integration: Connect with model makers (Mistral, Tech Mahindra), cloud providers (Amazon SageMaker), runtimes (LiteRT, ONNX Runtime), and ML Services (Dataloop, EyePop.ai).