Image Classification
PyTorch
foundation
android

SigLIP2: Optimized for Qualcomm Devices

SigLIP2 (Sigmoid Loss for Language-Image Pre-training 2) is a vision-language model from Google that computes cosine-similarity scores between images and text prompts. It can be used for zero-shot image classification, image search, and content moderation without any task-specific fine-tuning.

This is based on the implementation of SigLIP2 found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime Precision Chipset SDK Versions Download
ONNX float Universal QAIRT 2.45, ONNX Runtime 1.27.1 Download
ONNX w8a16 Universal QAIRT 2.45, ONNX Runtime 1.27.1 Download
QNN_DLC float Universal QAIRT 2.45 Download
QNN_DLC w8a16 Universal QAIRT 2.45 Download
TFLITE float Universal QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit SigLIP2 on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

  • Custom weights (e.g., fine-tuned checkpoints)
  • Custom input shapes
  • Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for SigLIP2 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

  • Image input resolution: 224x224
  • Model checkpoint: google/siglip2-base-patch16-224
  • Model size (image_encoder) (float): 352 MB
  • Model size (image_encoder) (w8a16): 92.8 MB
  • Model size (text_encoder) (float): 1.05 GB
  • Model size (text_encoder) (w8a16): 461 MB
  • Text sequence length: 64

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
image_encoder ONNX float Snapdragon® X2 Elite 4.737 ms 2 - 2 MB NPU
image_encoder ONNX float Snapdragon® X Elite 8.828 ms 182 - 182 MB NPU
image_encoder ONNX float Snapdragon® 8 Gen 3 Mobile 5.89 ms 0 - 309 MB NPU
image_encoder ONNX float Snapdragon® 8 Gen 1 Mobile 13.29 ms 1 - 293 MB NPU
image_encoder ONNX float Qualcomm® Dragonwing™ IQ-8275 11.992 ms 0 - 4 MB NPU
image_encoder ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 8.454 ms 0 - 194 MB NPU
image_encoder ONNX float Qualcomm® QCS8450 13.29 ms 1 - 293 MB NPU
image_encoder ONNX float Qualcomm® Dragonwing™ IQ-9075 12.863 ms 1 - 4 MB NPU
image_encoder ONNX float Qualcomm® Dragonwing™ IQ-X7181 8.828 ms 182 - 182 MB NPU
image_encoder ONNX float Qualcomm® Dragonwing™ Q-8750 4.244 ms 0 - 177 MB NPU
image_encoder ONNX float Snapdragon® 8 Elite Mobile 4.244 ms 0 - 177 MB NPU
image_encoder ONNX float Snapdragon® 8 Elite Gen 5 Mobile 3.235 ms 1 - 167 MB NPU
image_encoder ONNX w8a16 Snapdragon® X2 Elite 2.393 ms 1 - 1 MB NPU
image_encoder ONNX w8a16 Snapdragon® X Elite 5.555 ms 90 - 90 MB NPU
image_encoder ONNX w8a16 Snapdragon® 8 Gen 3 Mobile 3.664 ms 0 - 290 MB NPU
image_encoder ONNX w8a16 Snapdragon® 8 Gen 1 Mobile 7.586 ms 2 - 282 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ QCS6490 23.282 ms 2 - 5 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-8275 5.071 ms 0 - 4 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 5.293 ms 0 - 118 MB NPU
image_encoder ONNX w8a16 Qualcomm® QCS8450 7.586 ms 2 - 282 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-9075 5.26 ms 0 - 3 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-X7181 5.555 ms 90 - 90 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-6690 54.258 ms 0 - 398 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-7790 6.994 ms 0 - 270 MB NPU
image_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-8750 2.857 ms 0 - 236 MB NPU
image_encoder ONNX w8a16 Snapdragon® 8 Elite Mobile 2.857 ms 0 - 236 MB NPU
image_encoder ONNX w8a16 Snapdragon® 8 Elite Gen 5 Mobile 2.102 ms 0 - 239 MB NPU
image_encoder ONNX w8a16 Snapdragon® 7 Gen 4 Mobile 6.994 ms 0 - 270 MB NPU
image_encoder QNN_DLC float Snapdragon® X2 Elite 4.008 ms 1 - 1 MB NPU
image_encoder QNN_DLC float Snapdragon® X Elite 9.243 ms 1 - 1 MB NPU
image_encoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 6.033 ms 0 - 264 MB NPU
image_encoder QNN_DLC float Snapdragon® 8 Gen 1 Mobile 13.642 ms 0 - 242 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 12.113 ms 1 - 4 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 37.021 ms 1 - 174 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 8.55 ms 1 - 2 MB NPU
image_encoder QNN_DLC float Qualcomm® SA8775P 11.398 ms 1 - 175 MB NPU
image_encoder QNN_DLC float Qualcomm® SA8650P 11.398 ms 1 - 175 MB NPU
image_encoder QNN_DLC float Qualcomm® SA8255P 11.398 ms 1 - 175 MB NPU
image_encoder QNN_DLC float Qualcomm® QCS8450 13.642 ms 0 - 242 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 11.25 ms 1 - 3 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 9.243 ms 1 - 1 MB NPU
image_encoder QNN_DLC float Qualcomm® Dragonwing™ Q-8750 4.233 ms 1 - 176 MB NPU
image_encoder QNN_DLC float Qualcomm® SA7255P 37.021 ms 1 - 174 MB NPU
image_encoder QNN_DLC float Qualcomm® SA8295P 14.17 ms 1 - 150 MB NPU
image_encoder QNN_DLC float Snapdragon® 8 Elite Mobile 4.233 ms 1 - 176 MB NPU
image_encoder QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 3.401 ms 1 - 163 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® X2 Elite 4.048 ms 0 - 0 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® X Elite 9.173 ms 0 - 0 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® 8 Gen 3 Mobile 5.536 ms 0 - 313 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-8275 7.491 ms 0 - 3 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-8275 19.113 ms 0 - 253 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 8.427 ms 0 - 2 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® SA8775P 8.674 ms 0 - 255 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® SA8650P 8.674 ms 0 - 255 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® SA8255P 8.674 ms 0 - 255 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-9075 8.464 ms 2 - 4 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-X7181 9.173 ms 0 - 0 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-6690 92.066 ms 0 - 422 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-7790 13.912 ms 0 - 401 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-8750 4.467 ms 0 - 254 MB NPU
image_encoder QNN_DLC w8a16 Qualcomm® SA7255P 19.113 ms 0 - 253 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® 8 Elite Mobile 4.467 ms 0 - 254 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® 8 Elite Gen 5 Mobile 3.251 ms 0 - 256 MB NPU
image_encoder QNN_DLC w8a16 Snapdragon® 7 Gen 4 Mobile 13.912 ms 0 - 401 MB NPU
image_encoder TFLITE float Snapdragon® 8 Gen 3 Mobile 5.918 ms 0 - 271 MB NPU
image_encoder TFLITE float Snapdragon® 8 Gen 1 Mobile 13.259 ms 0 - 242 MB NPU
image_encoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 12.079 ms 0 - 184 MB NPU
image_encoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 36.75 ms 0 - 175 MB NPU
image_encoder TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 8.058 ms 0 - 2 MB NPU
image_encoder TFLITE float Qualcomm® SA8775P 11.282 ms 0 - 176 MB NPU
image_encoder TFLITE float Qualcomm® SA8650P 11.282 ms 0 - 176 MB NPU
image_encoder TFLITE float Qualcomm® SA8255P 11.282 ms 0 - 176 MB NPU
image_encoder TFLITE float Qualcomm® QCS8450 13.259 ms 0 - 242 MB NPU
image_encoder TFLITE float Qualcomm® Dragonwing™ IQ-9075 11.143 ms 0 - 184 MB NPU
image_encoder TFLITE float Qualcomm® Dragonwing™ Q-8750 4.162 ms 0 - 180 MB NPU
image_encoder TFLITE float Qualcomm® SA7255P 36.75 ms 0 - 175 MB NPU
image_encoder TFLITE float Qualcomm® SA8295P 13.929 ms 0 - 161 MB NPU
image_encoder TFLITE float Snapdragon® 8 Elite Mobile 4.162 ms 0 - 180 MB NPU
image_encoder TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 3.304 ms 0 - 169 MB NPU
text_encoder ONNX float Snapdragon® X2 Elite 1.483 ms 1 - 1 MB NPU
text_encoder ONNX float Snapdragon® X Elite 2.896 ms 542 - 542 MB NPU
text_encoder ONNX float Snapdragon® 8 Gen 3 Mobile 2.733 ms 0 - 253 MB NPU
text_encoder ONNX float Snapdragon® 8 Gen 1 Mobile 4.483 ms 0 - 244 MB NPU
text_encoder ONNX float Qualcomm® Dragonwing™ IQ-8275 3.768 ms 0 - 4 MB NPU
text_encoder ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 3.509 ms 0 - 643 MB NPU
text_encoder ONNX float Qualcomm® QCS8450 4.483 ms 0 - 244 MB NPU
text_encoder ONNX float Qualcomm® Dragonwing™ IQ-9075 3.51 ms 0 - 3 MB NPU
text_encoder ONNX float Qualcomm® Dragonwing™ IQ-X7181 2.896 ms 542 - 542 MB NPU
text_encoder ONNX float Qualcomm® Dragonwing™ Q-8750 2.446 ms 0 - 139 MB NPU
text_encoder ONNX float Snapdragon® 8 Elite Mobile 2.446 ms 0 - 139 MB NPU
text_encoder ONNX float Snapdragon® 8 Elite Gen 5 Mobile 2.4 ms 0 - 140 MB NPU
text_encoder ONNX w8a16 Snapdragon® X2 Elite 0.853 ms 1 - 1 MB NPU
text_encoder ONNX w8a16 Snapdragon® X Elite 1.716 ms 459 - 459 MB NPU
text_encoder ONNX w8a16 Snapdragon® 8 Gen 3 Mobile 1.452 ms 0 - 271 MB NPU
text_encoder ONNX w8a16 Snapdragon® 8 Gen 1 Mobile 2.801 ms 0 - 275 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ QCS6490 8.218 ms 0 - 3 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-8275 2.097 ms 0 - 3 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 1.859 ms 0 - 515 MB NPU
text_encoder ONNX w8a16 Qualcomm® QCS8450 2.801 ms 0 - 275 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-9075 2.194 ms 0 - 3 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ IQ-X7181 1.716 ms 459 - 459 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-6690 13.217 ms 0 - 259 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-7790 2.849 ms 0 - 239 MB NPU
text_encoder ONNX w8a16 Qualcomm® Dragonwing™ Q-8750 1.266 ms 0 - 233 MB NPU
text_encoder ONNX w8a16 Snapdragon® 8 Elite Mobile 1.266 ms 0 - 233 MB NPU
text_encoder ONNX w8a16 Snapdragon® 8 Elite Gen 5 Mobile 1.248 ms 0 - 233 MB NPU
text_encoder ONNX w8a16 Snapdragon® 7 Gen 4 Mobile 2.849 ms 0 - 239 MB NPU
text_encoder QNN_DLC float Snapdragon® X2 Elite 1.582 ms 0 - 0 MB NPU
text_encoder QNN_DLC float Snapdragon® X Elite 3.091 ms 0 - 0 MB NPU
text_encoder QNN_DLC float Snapdragon® 8 Gen 3 Mobile 2.696 ms 0 - 245 MB NPU
text_encoder QNN_DLC float Snapdragon® 8 Gen 1 Mobile 4.476 ms 0 - 238 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 3.75 ms 0 - 2 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 10.032 ms 0 - 133 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 3.496 ms 0 - 2 MB NPU
text_encoder QNN_DLC float Qualcomm® SA8775P 3.643 ms 0 - 134 MB NPU
text_encoder QNN_DLC float Qualcomm® SA8650P 3.643 ms 0 - 134 MB NPU
text_encoder QNN_DLC float Qualcomm® SA8255P 3.643 ms 0 - 134 MB NPU
text_encoder QNN_DLC float Qualcomm® QCS8450 4.476 ms 0 - 238 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 3.497 ms 0 - 2 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 3.091 ms 0 - 0 MB NPU
text_encoder QNN_DLC float Qualcomm® Dragonwing™ Q-8750 2.489 ms 0 - 135 MB NPU
text_encoder QNN_DLC float Qualcomm® SA7255P 10.032 ms 0 - 133 MB NPU
text_encoder QNN_DLC float Qualcomm® SA8295P 4.65 ms 0 - 122 MB NPU
text_encoder QNN_DLC float Snapdragon® 8 Elite Mobile 2.489 ms 0 - 135 MB NPU
text_encoder QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 2.388 ms 0 - 135 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® X2 Elite 1.047 ms 1 - 1 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® X Elite 1.941 ms 0 - 0 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® 8 Gen 3 Mobile 1.455 ms 0 - 261 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-8275 2.073 ms 0 - 2 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-8275 3.988 ms 0 - 226 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 1.885 ms 0 - 2 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® SA8775P 2.249 ms 0 - 227 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® SA8650P 2.249 ms 0 - 227 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® SA8255P 2.249 ms 0 - 227 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-9075 2.079 ms 0 - 2 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ IQ-X7181 1.941 ms 0 - 0 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-6690 14.42 ms 0 - 261 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-7790 2.825 ms 0 - 237 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® Dragonwing™ Q-8750 1.28 ms 0 - 227 MB NPU
text_encoder QNN_DLC w8a16 Qualcomm® SA7255P 3.988 ms 0 - 226 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® 8 Elite Mobile 1.28 ms 0 - 227 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® 8 Elite Gen 5 Mobile 1.257 ms 0 - 228 MB NPU
text_encoder QNN_DLC w8a16 Snapdragon® 7 Gen 4 Mobile 2.825 ms 0 - 237 MB NPU
text_encoder TFLITE float Snapdragon® 8 Gen 3 Mobile 2.729 ms 0 - 259 MB NPU
text_encoder TFLITE float Snapdragon® 8 Gen 1 Mobile 4.453 ms 0 - 241 MB NPU
text_encoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 3.749 ms 0 - 543 MB NPU
text_encoder TFLITE float Qualcomm® Dragonwing™ IQ-8275 10.083 ms 0 - 142 MB NPU
text_encoder TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 3.501 ms 0 - 3 MB NPU
text_encoder TFLITE float Qualcomm® SA8775P 3.676 ms 0 - 142 MB NPU
text_encoder TFLITE float Qualcomm® SA8650P 3.676 ms 0 - 142 MB NPU
text_encoder TFLITE float Qualcomm® SA8255P 3.676 ms 0 - 142 MB NPU
text_encoder TFLITE float Qualcomm® QCS8450 4.453 ms 0 - 241 MB NPU
text_encoder TFLITE float Qualcomm® Dragonwing™ IQ-9075 3.528 ms 0 - 543 MB NPU
text_encoder TFLITE float Qualcomm® Dragonwing™ Q-8750 2.41 ms 0 - 146 MB NPU
text_encoder TFLITE float Qualcomm® SA7255P 10.083 ms 0 - 142 MB NPU
text_encoder TFLITE float Qualcomm® SA8295P 4.685 ms 0 - 129 MB NPU
text_encoder TFLITE float Snapdragon® 8 Elite Mobile 2.41 ms 0 - 146 MB NPU
text_encoder TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 2.411 ms 0 - 147 MB NPU

License

  • The license for the original implementation of SigLIP2 can be found here.

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for qualcomm/SigLIP2