Qualcomm AI Hub

Qwen3-8B

Qwen3-8B is a flagship on-device language model built for complex reasoning and long-context tasks. It delivers cloud-level intelligence on edge devices with low latency and high efficiency.

Qwen3-Embedding

Qwen3-Embedding is a high-precision vector model for semantic search and RAG retrieval on edge. It provides fast, stable embeddings for multilingual understanding and document matching.

Qwen3-Reranker-0.6B

Qwen3-Reranker-0.6B is a lightweight reranking model that improves top-k retrieval quality in local RAG pipelines. It boosts answer accuracy by refining relevance with minimal memory and power cost.

Qwen2.5-VL-3B

Qwen2.5-VL-3B is a breakthrough vision language model (VLM), marking a transition from pure text interaction to multimodal perception on edge devices.

Qwen2-7B-SSD

Qwen2-7B-SSD is a flagship edge-side model with comprehensive performance benchmark, built on Alibaba powerful Qwen2 architecture.

Llama 3.2 3B

Meta Llama 3.2 3B model is a lightweight and efficient model designed for mobile devices with limited resources.

Llama 3.1 8B

Llama 3.1 8B model finds a golden balance between performance and efficiency, making it the most popular and versatile edge-side model.

BGE-Base-Zh-V1.5

BGE-Base-Zh-V1.5 is a lightweight Chinese general semantic embedding model developed by BAAI, essential for building edge RAG applications.

AOT-GAN

High resolution image in-painting on-device.

Baichuan2-7B

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Beit

Imagenet classifier and general purpose backbone.

BGNet

Segment images in real-time on device.

BiseNet

Segment images or video by class in real-time on device.

Conditional-DETR-ResNet50

Transformer based object detector with ResNet50 backbone.

ControlNet-Canny

Generating visual arts from text prompt and input guiding image.

ConvNext-Base

Imagenet classifier and general purpose backbone.

ConvNext-Tiny

Imagenet classifier and general purpose backbone.

DDRNet23-Slim

Segment images or video by class in real-time on device.

3D-Deep-BOX

Real-time 3D object detection.

DeepLabXception

Deep Convolutional Neural Network model for semantic segmentation.

DeepLabV3-Plus-MobileNet

Deep Convolutional Neural Network model for semantic segmentation.

DeepLabV3-ResNet50

Deep Convolutional Neural Network model for semantic segmentation.

DenseNet-121

Imagenet classifier and general purpose backbone.

Depth-Anything

Deep Convolutional Neural Network model for depth estimation.

Depth-Anything-V2

Deep Convolutional Neural Network model for depth estimation.

DETR-ResNet101

Transformer based object detector with ResNet101 backbone.

DETR-ResNet101-DC5

Transformer based object detector with ResNet101 backbone (dilated C5 stage).

DETR-ResNet50

Transformer based object detector with ResNet50 backbone.

DETR-ResNet50-DC5

Transformer based object detector with ResNet50 backbone (dilated C5 stage).

DLA-102-X

Imagenet classifier and general purpose backbone.

EasyOCR

Ready-to-use OCR with 80+ supported languages and all popular writing scripts.

EfficientFormer

Imagenet classifier and general purpose backbone.

EfficientNet-B0

Imagenet classifier and general purpose backbone.

EfficientNet-B4

Imagenet classifier and general purpose backbone.

EfficientNet-V2-s

Imagenet classifier and general purpose backbone.

EfficientViT-b2-cls

Imagenet classifier and general purpose backbone.

EfficientViT-l2-cls

Imagenet classifier and general purpose backbone.

EfficientViT-l2-seg

Semantic segmentation in higher resuolution.

ESRGAN

Upscale images and remove image noise.

Facial-Attribute-Detection

Comprehensive facial analysis by extracting face features.

Lightweight-Face-Detection

Lightweight and efficient face detector.

Facial-Landmark-Detection

Real-time 3D facial landmark detection optimized for mobile and edge.

FastSam-S

Generate high quality segmentation mask on device.

FastSam-X

Generate high quality segmentation mask on device.

FCN-ResNet50

Fully-convolutional network model for image segmentation.

FFNet-122NS-LowRes

Semantic segmentation for automotive street scenes.

FFNet-40S

Semantic segmentation for automotive street scenes.

FFNet-54S

Semantic segmentation for automotive street scenes.

FFNet-78S

Semantic segmentation for automotive street scenes.

FFNet-78S-LowRes

Semantic segmentation for automotive street scenes.

First-Order-Motion-Model

Animation of Still Image from Source Video.

Person-Foot-Detection

Multi-task Human detector.

PPE-Detection

Object detection for personal protective equipments (PPE).

GoogLeNet

Imagenet classifier and general purpose backbone.

HRNetPose

Perform accurate human pose estimation.

HRNet-W48-OCR

Semantic segmentation in higher resolution.

HuggingFace-WavLM-Base-Plus

Real-time Speech processing.

IBM-Granite-v3.1-8B-Instruct

State-of-the-art large language model useful on a variety of code understanding and generation tasks.

Inception-v3

Imagenet classifier and general purpose backbone.

IndusQ-1.1B

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

JAIS-6p7b-Chat

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

LaMa-Dilated

High resolution image in-painting on-device.

LeViT

Imagenet classifier and general purpose backbone.

LiteHRNet

Human pose estimation.

Llama-v2-7B-Chat

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Llama-v3.1-8B-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Llama-SEA-LION-v3.5-8B-R

Llama-SEA-LION-v3.5-8B-R is a hybrid model offering versatile functionality, handling both complex reasoning tasks and general text generation, with mode selection managed through the tokenizer's chat template.

Llama-v3.2-1B-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Llama-v3.2-3B-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Llama-v3-8B-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Llama3-TAIDE-LX-8B-Chat-Alpha1

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Mask2Former

Real-time object segmentation.

MediaPipe-Face-Detection

Detect faces and locate facial features in real-time video and image streams.

MediaPipe-Hand-Detection

Real-time hand detection optimized for mobile and edge.

MediaPipe-Pose-Estimation

Detect and track human face, hand, and torso in real-time images and video streams.

MediaPipe-Selfie-Segmentation

Segments the person from background in a selfie image and realtime background segmentation in video conferencing.

Midas-V2

Deep Convolutional Neural Network model for depth estimation.

Ministral-3B

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Mistral-3B

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Mistral-7B-Instruct-v0.3

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

MNASNet05

Imagenet classifier and general purpose backbone.

Mobile-VIT

Imagenet classifier and general purpose backbone.

MobileNet-v2

Imagenet classifier and general purpose backbone.

MobileNet-v3-Large

Imagenet classifier and general purpose backbone.

MobileNet-v3-Small

Imagenet classifier and general purpose backbone.

MobileSam

Faster Segment Anything: Towards lightweight SAM for mobile applications.

Movenet

Perform accurate human pose estimation.

NASNet

Imagenet classifier and general purpose backbone.

Nomic-Embed-Text

Resizable Production Embeddings.

OpenAI-Clip

Multi-modal foundational model for vision and language tasks like image/text similarity and for zero-shot image classification.

Phi-3.5-Mini-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

PidNet

Segment images or video by class in real-time on device.

PLaMo-1B

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Posenet-Mobilenet

Perform accurate human pose estimation.

QuickSRNetLarge

Upscale images and remove image noise.

QuickSRNetMedium

Upscale images and remove image noise.

QuickSRNetSmall

Upscale images and remove image noise.

Qwen2-7B-Instruct

State-of-the-art large language model useful on a variety of language understanding and generation tasks.

Real-ESRGAN-General-x4v3

Upscale images and remove image noise.

Real-ESRGAN-x4plus

Upscale images and remove image noise.

RegNet

Imagenet classifier and general purpose backbone.

ResNet101

Imagenet classifier and general purpose backbone.

ResNet18

Imagenet classifier and general purpose backbone.

ResNet50

Imagenet classifier and general purpose backbone.

ResNet-2Plus1D

Sports and human action recognition in videos.

ResNet-3D

Sports and human action recognition in videos.

ResNet-Mixed-Convolution

Sports and human action recognition in videos.

ResNeXt101

Imagenet classifier and general purpose backbone.

ResNeXt50

Imagenet classifier and general purpose backbone.

RF-DETR

Transformer based object detection model architecture developed by Roboflow.

RTMDet

Real-time object detection optimized for mobile and edge.

RTMPose-Body2d

Human pose estimation.

SalsaNext

Semantic segmentation model optimized for LiDAR point cloud data.

Segment-Anything-Model-2

High-quality segmentation in images and videos with real-time performance and minimal user interaction.

Segformer-Base

Real-time object segmentation.

SESR-M5

Upscale images in real time.

Shufflenet-v2

Imagenet classifier and general purpose backbone.

Simple-Bev

Construct a bird's eye view from sensors mounted on a vehicle.

SINet

Lightweight portrait segmentation for background removal.

SqueezeNet-1.1

Imagenet classifier and general purpose backbone.

Stable-Diffusion-v1.5

State-of-the-art generative AI model used to generate detailed images conditioned on text descriptions.

Stable-Diffusion-v2.1

State-of-the-art generative AI model used to generate detailed images conditioned on text descriptions.

Swin-Base

Imagenet classifier and general purpose backbone.

Swin-Small

Imagenet classifier and general purpose backbone.

Swin-Tiny

Imagenet classifier and general purpose backbone.

TrOCR

Transformer based model for state-of-the-art optical character recognition (OCR) on both printed and handwritten text.

Unet-Segmentation

Real-time segmentation optimized for mobile and edge.

Video-MAE

Sports and human action recognition in videos.

VIT

Imagenet classifier and general purpose backbone.

Whisper-Base-En

Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace.

Whisper-Large-V3-Turbo

Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace.

Whisper-Small-En

Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace.

Whisper-Tiny-En

Transformer-based automatic speech recognition (ASR) model for multilingual transcription and translation available on HuggingFace.

WideResNet50

Imagenet classifier and general purpose backbone.

XLSR

Upscale images in real time.

YamNet

Audio Event classification Model.

YOLOv10-Detection

Real-time object detection optimized for mobile and edge by Ultralytics.

YOLOv11-Detection

Real-time object detection optimized for mobile and edge by Ultralytics.

YOLOv11-Segmentation

Real-time object segmentation optimized for mobile and edge by Ultralytics.

Yolo-v3

Real-time object detection optimized for mobile and edge.

Yolo-v5

Real-time object detection optimized for mobile and edge.

Yolo-v6

Real-time object detection optimized for mobile and edge.

Yolo-v7

Real-time object detection optimized for mobile and edge.

YOLOv8-Detection

Real-time object detection optimized for mobile and edge by Ultralytics.

YOLOv8-Segmentation

Real-time object segmentation optimized for mobile and edge by Ultralytics.

Yolo-X

Real-time object detection optimized for mobile and edge.

Tags

Qwen3-8B

Qwen3-Embedding

Qwen3-Reranker-0.6B

Qwen2.5-VL-3B

Qwen2-7B-SSD

Llama 3.2 3B

Llama 3.1 8B

BGE-Base-Zh-V1.5

AOT-GAN

Baichuan2-7B

Beit

BGNet

BiseNet

Conditional-DETR-ResNet50

ControlNet-Canny

ConvNext-Base

ConvNext-Tiny

DDRNet23-Slim

3D-Deep-BOX

DeepLabXception

DeepLabV3-Plus-MobileNet

DeepLabV3-ResNet50

DenseNet-121

Depth-Anything

Depth-Anything-V2

DETR-ResNet101

DETR-ResNet101-DC5

DETR-ResNet50

DETR-ResNet50-DC5

DLA-102-X

EasyOCR

EfficientFormer

EfficientNet-B0

EfficientNet-B4

EfficientNet-V2-s

EfficientViT-b2-cls

EfficientViT-l2-cls

EfficientViT-l2-seg

ESRGAN

Facial-Attribute-Detection

Lightweight-Face-Detection

Facial-Landmark-Detection

FastSam-S

FastSam-X

FCN-ResNet50

FFNet-122NS-LowRes

FFNet-40S

FFNet-54S

FFNet-78S

FFNet-78S-LowRes

First-Order-Motion-Model

Person-Foot-Detection

PPE-Detection

GoogLeNet

HRNetPose

HRNet-W48-OCR

HuggingFace-WavLM-Base-Plus

IBM-Granite-v3.1-8B-Instruct

Inception-v3

IndusQ-1.1B

JAIS-6p7b-Chat

LaMa-Dilated

LeViT

LiteHRNet

Llama-v2-7B-Chat

Llama-v3.1-8B-Instruct

Llama-SEA-LION-v3.5-8B-R

Llama-v3.2-1B-Instruct

Llama-v3.2-3B-Instruct

Llama-v3-8B-Instruct

Llama3-TAIDE-LX-8B-Chat-Alpha1

Mask2Former

MediaPipe-Face-Detection

MediaPipe-Hand-Detection

MediaPipe-Pose-Estimation

MediaPipe-Selfie-Segmentation

Midas-V2

Ministral-3B

Mistral-3B