Q

GenieX

Developer tool

by Qualcomm

Open-source on-device inference runtime that runs language and vision-language models locally on Snapdragon NPU, GPU and CPU through a CLI, Python and Android SDKs, Docker and an OpenAI-compatible local server.

Find alternatives to GenieX

What it does

• Model inference

Features 5

• CLIRuns and serves models straight from the terminal.
• Python SDKEmbeds inference in applications with the Python SDK.
• Android SDKJava/Kotlin SDK for on-device mobile apps.
• Docker imageContainerized image for reproducible deployments.
• OpenAI-compatible serverDrop-in local server for existing OpenAI clients.

Sources

Pricing

Official documentation · 17 Oct 2026

Legal section of the GenieX docs: "GenieX is released under the BSD 3-Clause License." (the GitHub repository qualcomm/GenieX shows the same licence). Standard 4.1 row 1 (an open licence) -> open-source.

Description

Official documentation · 17 Oct 2026

"GenieX is an on-device Gen AI inference runtime built for Qualcomm platforms. It is the easiest way to run frontier language and vision-language models locally on Hexagon NPU, Adreno GPU, or CPU with a few lines of code." GitHub README: "One C SDK underneath is exposed through a CLI, Python, Kotlin/Java, Docker, and an OpenAI-compatible server."

What it does

Official documentation · 17 Oct 2026

"run frontier language and vision-language models locally" (inference runtime; no training or hosting claim).

Sold within

Official documentation · 17 Oct 2026

Released under the BSD 3-Clause License with an install-and-run quickstart (CLI, Python, Android from Maven Central, Docker); no purchase step is described, so it stands alone.

Name

Official documentation · 17 Oct 2026

Page title 'What is GenieX - Qualcomm AI Hub'; the Qualcomm developer blog of 2026-06 is titled 'Run Gen AI on Qualcomm chipsets with GenieX (formerly NexaSDK)'.

URL

Official documentation · 17 Oct 2026

Qualcomm AI Hub's GenieX documentation, page 'What is GenieX', read 2026-10-17.

Deployment

Platform listing · 17 Oct 2026

"Run frontier LLMs and VLMs locally on Qualcomm devices"; "GenieX runs only on Qualcomm Snapdragon" (Windows ARM64, Android, Linux ARM64).

Interfaces

Platform listing · 17 Oct 2026

README: 'One C SDK underneath is exposed through a CLI, Python, Kotlin/Java, Docker, and an OpenAI-compatible server.'

Intended for

Official documentation · 17 Oct 2026

Developer documentation: CLI, Python SDK, Android SDK, Docker, tutorials and benchmarking.

Also from Qualcomm

• Qualcomm AI Hub Developer tool

  A developer platform for optimising, benchmarking and deploying on-device AI models against 300+ pre-optimised models and 50+ real Qualcomm devices in the cloud.

• Qualcomm AI Inference Suite Platform

  A cloud and on-premises inference platform offering an SDK, OpenAI-compatible APIs and ready-made chat, RAG, image and code applications on top of Qualcomm's AI accelerators.

• Qualcomm Cloud AI 100 Ultra Hardware

  A PCIe AI inference accelerator card with up to 870 TOPS INT8 and 128GB of memory, built for generative AI and LLM inference in the data centre.

• Qualcomm Dragonfly AI200 Hardware

  Rack-scale data-centre inference accelerator carrying 768 GB of LPDDR per card, announced October 2025 with commercial availability expected in 2026.

• Qualcomm Dragonfly AI250 Hardware

  A second-generation rack-scale AI inference platform using Qualcomm High Bandwidth Compute (HBC) Gen 1 for memory-bound agentic inference.

• Qualcomm Dragonfly AI300 Hardware

  A third-generation rack-scale AI inference platform using Qualcomm High Bandwidth Compute (HBC) Gen 2 for hyperscale LLM and agentic deployments.

• Snapdragon Hardware

  Mobile and PC processor family whose on-device AI engine runs models locally on phones, laptops and cars.

• Snapdragon X Series Hardware

  A family of Snapdragon compute platforms for Windows PCs (X, X2 Elite, X2 Plus) sold on NPU-accelerated on-device AI performance.

• Qualcomm Dragonwing IQ-2390 Hardware

  Embedded edge-AI processor in early access; part of Qualcomm's Dragonwing IQ2 series for IoT.

• Qualcomm Dragonwing IQ9 Series Hardware

  Embedded IoT/edge-AI processor family, distinct from the Dragonfly AI200/250/300 data-centre accelerator line; an IQ-9075 Evaluation Kit is purchasable through Qualcomm's developer hardware section.

Open Qualcomm in the directory

Something wrong here? Send a correction — quote this product id: qualcomm-geniex.