OpenAI-Clip: Optimized for Qualcomm Devices

Contrastive Language-Image Pre-Training (CLIP) uses a ViT like transformer to get visual features and a causal language model to get the text features. Both the text and visual features can then be used for a variety of zero-shot learning tasks.

This is based on the implementation of OpenAI-Clip 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 OpenAI-Clip 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 OpenAI-Clip on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Image input resolution: 224x224
• Model checkpoint: ViT-B/16
• Model size (float): 571 MB
• Number of parameters: 150M
• Text context length: 77

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
OpenAI-Clip	ONNX	float	Snapdragon® X2 Elite	19.637 ms	2 - 2 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® X Elite	33.095 ms	294 - 294 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 3 Mobile	21.437 ms	0 - 1183 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 1 Mobile	47.733 ms	1 - 894 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	36.198 ms	0 - 4 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	31.767 ms	0 - 322 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS8450	47.733 ms	1 - 894 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	36.21 ms	1 - 4 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	33.095 ms	294 - 294 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ Q-8750	17.014 ms	1 - 958 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Elite Mobile	17.014 ms	1 - 958 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	13.46 ms	1 - 806 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® X2 Elite	13.461 ms	1 - 1 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® X Elite	27.11 ms	174 - 174 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	18.801 ms	0 - 750 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	41.504 ms	0 - 756 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	145.168 ms	1 - 4 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	24.999 ms	0 - 4 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	26.429 ms	0 - 3 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® QCS8450	41.504 ms	0 - 756 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	25.82 ms	0 - 3 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	27.11 ms	174 - 174 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	300.91 ms	0 - 664 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	37.93 ms	1 - 669 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	14.951 ms	0 - 640 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Elite Mobile	14.951 ms	0 - 640 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	12.875 ms	0 - 681 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	37.93 ms	1 - 669 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X2 Elite	14.695 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X Elite	33.278 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	21.892 ms	0 - 924 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	47.136 ms	0 - 854 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	36.327 ms	1 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	99.093 ms	1 - 704 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	31.769 ms	1 - 2 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8775P	36.663 ms	1 - 701 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8650P	36.663 ms	1 - 701 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8255P	36.663 ms	1 - 701 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8450	47.136 ms	0 - 854 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	36.799 ms	1 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	33.278 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	17.173 ms	0 - 694 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA7255P	99.093 ms	1 - 704 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8295P	38.482 ms	1 - 797 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite Mobile	17.173 ms	0 - 694 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	13.361 ms	0 - 703 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® X2 Elite	14.519 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® X Elite	33.329 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	22.998 ms	0 - 748 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	30.238 ms	0 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	67.744 ms	1 - 604 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	32.996 ms	0 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8775P	32.428 ms	0 - 603 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8650P	32.428 ms	0 - 603 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8255P	32.428 ms	0 - 603 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	32.65 ms	0 - 2 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	33.329 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	464.549 ms	1 - 687 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	41.294 ms	0 - 658 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	18.275 ms	0 - 633 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA7255P	67.744 ms	1 - 604 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	18.275 ms	0 - 633 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	14.484 ms	0 - 677 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	41.294 ms	0 - 658 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	21.683 ms	0 - 898 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	40.539 ms	0 - 907 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	35.295 ms	0 - 297 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	95.572 ms	0 - 664 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	31.456 ms	0 - 3 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8775P	36.013 ms	0 - 671 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8650P	36.013 ms	0 - 671 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8255P	36.013 ms	0 - 671 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8450	40.539 ms	0 - 907 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	36.27 ms	0 - 296 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	17.453 ms	0 - 698 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA7255P	95.572 ms	0 - 664 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8295P	38.351 ms	0 - 622 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite Mobile	17.453 ms	0 - 698 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	13.037 ms	0 - 537 MB	NPU

License

• The license for the original implementation of OpenAI-Clip can be found here.

References

• Learning Transferable Visual Models From Natural Language Supervision
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.