v0.48.0

See https://github.com/qualcomm/ai-hub-models/releases/v0.48.0 for changelog.

LICENSE

@@ -0,0 +1 @@
+The license of the original trained model can be found at https://huggingface.co/Qwen/Qwen3-4B/blob/main/LICENSE.

README.md

@@ -0,0 +1,110 @@
+---
+library_name: pytorch
+license: other
+tags:
+- llm
+- generative_ai
+- android
+pipeline_tag: text-generation
+
+---
+
+![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/web-assets/model_demo.png)
+
+# Qwen3-4B: Optimized for Qualcomm Devices
+
+The Qwen3-4B is a state-of-the-art multilingual base language model with 4 billion parameters, excelling in language understanding, generation, coding, and mathematics.
+
+This is based on the implementation of Qwen3-4B found [here](https://huggingface.co/Qwen/Qwen3-4B).
+This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/qwen3_4b) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
+
+Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
+
+## Deploying Qwen3-4B on-device
+
+Please follow the [LLM on-device deployment](https://github.com/quic/ai-hub-apps/tree/main/tutorials/llm_on_genie) tutorial.
+
+## 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 |
+|---|---|---|---|---|
+| GENIE | w4a16 | Snapdragon® 8 Elite Mobile | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_snapdragon_8_elite.zip)
+| GENIE | w4a16 | Snapdragon® X2 Elite | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_snapdragon_x2_elite.zip)
+| GENIE | w4a16 | Snapdragon® X Elite | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_snapdragon_x_elite.zip)
+| GENIE | w4a16 | qualcomm-qcs8275 | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_qcs8275.zip)
+| GENIE | w4a16 | Qualcomm® QCS9075 | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_qcs9075.zip)
+| GENIE | w4a16 | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.42 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen3_4b/releases/v0.48.0/qwen3_4b-genie-w4a16-qualcomm_snapdragon_8_elite_gen5.zip)
+
+For more device-specific assets and performance metrics, visit **[Qwen3-4B on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/qwen3_4b)**.
+
+
+### Option 2: Export with Custom Configurations
+
+Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/qwen3_4b) 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 [Qwen3-4B on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/qwen3_4b) for usage instructions.
+
+## Model Details
+
+**Model Type:** Model_use_case.text_generation
+
+**Model Stats:**
+- Input sequence length for Prompt Processor: 128
+- Context lengths: 512,1024,2048,3072,4096
+- Use: Initiate conversation with prompt-processor and then token generator for subsequent iterations.
+- Minimum QNN SDK version required: 2.42.0
+- Supported languages: 100+ languages and dialects.
+- TTFT: Time To First Token is the time it takes to generate the first response token. This is expressed as a range because it varies based on the length of the prompt. The lower bound is for a short prompt (up to 128 tokens, i.e., one iteration of the prompt processor) and the upper bound is for a prompt using the full context length (4096 tokens).
+- Response Rate: Rate of response generation after the first response token. Measured on a short prompt with a long response (with thinking); may slow down when using longer context lengths.
+
+## Performance Summary
+| Model | Runtime | Precision | Chipset | Context Length | Response Rate (tokens per second) | Time To First Token (range, seconds)
+|---|---|---|---|---|---|---
+| Qwen3-4B | GENIE | w4a16 | Snapdragon® X2 Elite | 4096 | 29.989173316955565 | 0.07306119999999999 - 2.3379583999999998
+| Qwen3-4B | GENIE | w4a16 | Snapdragon® X Elite | 4096 | 14.786919736862183 | 0.1217166 - 3.8949312
+| Qwen3-4B | GENIE | w4a16 | Snapdragon® 8 Elite For Galaxy Mobile | 4096 | 25.431722831726074 | 0.0689817 - 2.2074144
+| Qwen3-4B | GENIE | w4a16 | Snapdragon® 8 Elite Gen 5 Mobile | 4096 | 29.032558250427247 | 0.0536603 - 1.7171296
+| Qwen3-4B | GENIE | w4a16 | Qualcomm® QCS9075 | 4096 | 12.981898975372314 | 0.1621608 - 5.1891456
+
+## License
+* The license for the original implementation of Qwen3-4B can be found
+  [here](https://huggingface.co/Qwen/Qwen3-4B/blob/main/LICENSE).
+
+## References
+* [Qwen3 Technical Report](https://arxiv.org/abs/2505.09388)
+* [Source Model Implementation](https://huggingface.co/Qwen/Qwen3-4B)
+
+## Community
+* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
+* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
+
+## Usage and Limitations
+
+This model may not be used for or in connection with any of the following applications:
+
+- Accessing essential private and public services and benefits;
+- Administration of justice and democratic processes;
+- Assessing or recognizing the emotional state of a person;
+- Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
+- Education and vocational training;
+- Employment and workers management;
+- Exploitation of the vulnerabilities of persons resulting in harmful behavior;
+- General purpose social scoring;
+- Law enforcement;
+- Management and operation of critical infrastructure;
+- Migration, asylum and border control management;
+- Predictive policing;
+- Real-time remote biometric identification in public spaces;
+- Recommender systems of social media platforms;
+- Scraping of facial images (from the internet or otherwise); and/or
+- Subliminal manipulation