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 2.50, ONNX Runtime 1.27.1	Download
ONNX	w8a16	Universal	QAIRT 2.50, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
QNN_DLC	w8a16	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	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® 8 Elite Gen 5 For Galaxy Mobile	13.245 ms	1 - 613 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	16.662 ms	0 - 602 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® X2 Elite	13.948 ms	2 - 2 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® X Elite	32.635 ms	294 - 294 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 3 Mobile	21.83 ms	1 - 992 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 1 Mobile	41.33 ms	1 - 848 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	35.502 ms	0 - 5 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	31.987 ms	0 - 322 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS8450	41.33 ms	1 - 848 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	36.392 ms	1 - 5 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	32.635 ms	294 - 294 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® Dragonwing™ Q-8750	16.662 ms	0 - 602 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	11.485 ms	0 - 658 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	14.918 ms	0 - 620 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® X2 Elite	11.929 ms	1 - 1 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® X Elite	25.222 ms	173 - 173 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	17.896 ms	0 - 739 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	34.933 ms	0 - 740 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	103.718 ms	3 - 6 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	23.545 ms	0 - 5 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	24.639 ms	0 - 195 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® QCS8450	34.933 ms	0 - 740 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	24.63 ms	0 - 4 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	25.222 ms	173 - 173 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	334.61 ms	0 - 690 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	37.089 ms	1 - 638 MB	NPU
OpenAI-Clip	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	14.918 ms	0 - 620 MB	NPU
OpenAI-Clip	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	37.089 ms	1 - 638 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	13.098 ms	0 - 489 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	16.747 ms	1 - 483 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X2 Elite	14.143 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X Elite	33.128 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	21.394 ms	0 - 688 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	37.296 ms	1 - 815 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	35.973 ms	1 - 4 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	31.39 ms	1 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8650P	36.675 ms	1 - 700 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8255P	36.675 ms	1 - 700 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8450	37.296 ms	1 - 815 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	36.419 ms	3 - 5 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	33.128 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	16.747 ms	1 - 483 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8295P	37.723 ms	1 - 591 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	11.506 ms	0 - 628 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	15.226 ms	0 - 591 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® X2 Elite	12.1 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® X Elite	28.154 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	18.724 ms	0 - 702 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	24.827 ms	0 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	27.141 ms	0 - 440 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8775P	26.821 ms	0 - 566 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8650P	26.821 ms	0 - 566 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA8255P	26.821 ms	0 - 566 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	27.154 ms	0 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	28.154 ms	0 - 0 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	432.665 ms	1 - 693 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	37.332 ms	0 - 611 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	15.226 ms	0 - 591 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Qualcomm® SA7255P	58.663 ms	0 - 564 MB	NPU
OpenAI-Clip	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	37.332 ms	0 - 611 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	12.822 ms	0 - 499 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	17.003 ms	0 - 490 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	21.13 ms	0 - 693 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	36.247 ms	0 - 677 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	34.563 ms	0 - 297 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	30.974 ms	0 - 3 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8650P	35.976 ms	0 - 670 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8255P	35.976 ms	0 - 670 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8450	36.247 ms	0 - 677 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	35.657 ms	0 - 297 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	17.003 ms	0 - 490 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA7255P	748.41 ms	11 - 35 MB	CPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8295P	37.294 ms	0 - 441 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.