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 1.25.0	Download
QNN_DLC	float	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
OpenAI-Clip	ONNX	float	Snapdragon® X2 Elite	10.356 ms	180 - 180 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® X Elite	22.143 ms	294 - 294 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 3 Mobile	14.961 ms	1 - 1096 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Gen 1 Mobile	26.596 ms	1 - 674 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS8550 (Proxy)	21.472 ms	0 - 323 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS8450	26.596 ms	1 - 674 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Elite Mobile	11.719 ms	1 - 870 MB	NPU
OpenAI-Clip	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	9.258 ms	1 - 578 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS9075	26.356 ms	1 - 46 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS8750	11.719 ms	1 - 870 MB	NPU
OpenAI-Clip	ONNX	float	Qualcomm® QCS7181	22.143 ms	294 - 294 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X2 Elite	9.648 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® X Elite	21.459 ms	1 - 1 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	14.445 ms	0 - 768 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	24.142 ms	0 - 609 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8275	59.355 ms	1 - 570 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	20.186 ms	1 - 3 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8450	24.142 ms	0 - 609 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite Mobile	11.311 ms	1 - 587 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA8295P	24.546 ms	1 - 514 MB	NPU
OpenAI-Clip	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	8.749 ms	0 - 419 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® SA7255P	59.355 ms	1 - 570 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS9075	23.86 ms	3 - 5 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS8750	11.311 ms	1 - 587 MB	NPU
OpenAI-Clip	QNN_DLC	float	Qualcomm® QCS7181	21.459 ms	1 - 1 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	14.25 ms	0 - 775 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	24.342 ms	0 - 617 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8275	59.423 ms	0 - 566 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8550 (Proxy)	20.105 ms	0 - 3 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8775P	333.229 ms	6 - 30 MB	CPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8650P	333.229 ms	6 - 30 MB	CPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8255P	333.229 ms	6 - 30 MB	CPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8450	24.342 ms	0 - 617 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite Mobile	11.342 ms	0 - 606 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA8295P	24.559 ms	0 - 518 MB	NPU
OpenAI-Clip	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	8.74 ms	0 - 433 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® SA7255P	59.423 ms	0 - 566 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS9075	23.106 ms	0 - 296 MB	NPU
OpenAI-Clip	TFLITE	float	Qualcomm® QCS8750	11.342 ms	0 - 606 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.