Hugging Face's logo

qualcomm
/
Qwen3-4B-Instruct-2507

Text Generation
PyTorch
llm
generative_ai
android
Model card Files
xet
 Community

Qwen3-4B-Instruct-2507: Optimized for Qualcomm Devices

The Qwen3-4B-Instruct-2507 is a state-of-the-art multilingual instruct language model with 4 billion parameters, excelling in language understanding, generation, coding, and mathematics. Unlike the base Qwen3-4B, this variant has only instruct training without thinking mode support.

This is based on the implementation of Qwen3-4B-Instruct-2507 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.

Deploying Qwen3-4B-Instruct-2507 on-device

Follow the GenieX quickstart to install GenieX and deploy the model on a target device.

See the LLM-on-Genie tutorial to run with the Genie runtime. Note: Genie support will be deprecated soon.

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 Gen 5 For Galaxy Mobile QAIRT 2.45 Download
GENIE w4a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.45 Download
GENIE w4a16 Snapdragon® X2 Elite QAIRT 2.45 Download
GENIE w4a16 Snapdragon® X Elite QAIRT 2.45 Download
GENIE w4a16 Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.45 Download
GENIE w4a16 Qualcomm® SA8775P QAIRT 2.45 Download
GENIE w4a16 Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.45 Download
GENIEX_LLAMACPP q4_0 Universal Download
GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® X2 Elite QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® X Elite QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit Qwen3-4B-Instruct-2507 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 Qwen3-4B-Instruct-2507 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

  • Response Rate: Rate of response generation after the first response token. Measured on a short prompt with a long response; may slow down when using longer context lengths.
  • 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).

Performance Summary

Model Runtime Precision Chipset Context Length Response Rate (tokens per second) Time To First Token (range, seconds)
Qwen3-4B-Instruct-2507 GENIE w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 27.985336303710938 0.05547 - 1.77504
Qwen3-4B-Instruct-2507 GENIE w4a16 Snapdragon® 8 Elite For Galaxy Mobile 4096 26.66240119934082 0.065065 - 2.08208
Qwen3-4B-Instruct-2507 GENIE w4a16 Snapdragon® X2 Elite 4096 32.163973808288574 0.07509 - 2.40288
Qwen3-4B-Instruct-2507 GENIE w4a16 Snapdragon® X Elite 4096 8.30799388885498 0.10081799999999999 - 3.2261759999999997
Qwen3-4B-Instruct-2507 GENIE w4a16 Qualcomm® Dragonwing™ IQ-8275 4096 18.114301681518555 0.094182 - 3.013824
Qwen3-4B-Instruct-2507 GENIE w4a16 Qualcomm® Dragonwing™ IQ-9075 4096 19.92993297576904 0.10179260000000001 - 3.2573632000000003
Qwen3-4B-Instruct-2507 GENIE w4a16 Qualcomm® Dragonwing™ IQ-X7181 4096 8.30799388885498 0.10081799999999999 - 3.2261759999999997
Qwen3-4B-Instruct-2507 GENIE w4a16 Qualcomm® Dragonwing™ Q-8750 4096 26.66240119934082 0.065065 - 2.08208
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 19.697929 1.3546813333333332 - 5.418725333333333
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 21.139921 0.4971393333333333 - 1.9885573333333333
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 22.245684 0.10473666666666664 - 0.4189466666666666
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 6.223475 2.981126161290322 - 95.39603716129031
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 12.660195 1.1085383548387098 - 35.47322735483871
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 11.399435 0.13752932258064515 - 4.400938322580645
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 21.303923 1.157875 - 4.6315
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 20.209841 0.5405866666666665 - 2.162346666666666
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 17.073282 0.12675333333333333 - 0.5070133333333333
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 4.981844 3.0638862258064514 - 98.04435922580645
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 10.070317 1.4112917741935482 - 45.16133677419354
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 8.015922 0.2074222580645161 - 6.637512258064516
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 36.120735 0.2838346666666666 - 1.1353386666666665
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 20.878213 0.2478723333333333 - 0.9914893333333332
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 20.247206 0.09196233333333334 - 0.36784933333333336
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 21.111574 0.5152424838709677 - 16.487759483870967
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 16.904976 0.48492470967741935 - 15.51759070967742
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 13.051477 0.12212519354838709 - 3.908006193548387
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 28.265948 0.6946986666666667 - 2.7787946666666667
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 24.221611 0.5377303333333333 - 2.1509213333333332
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 12.497374 0.22211166666666668 - 0.8884466666666667
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 7.389816 1.344167612903226 - 43.013363612903234
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 13.675057 1.2736976451612902 - 40.75832464516129
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 7.03239 0.3106614193548387 - 9.941165419354839
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 3.893702 15.624039 - 62.496156
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 2.531282 22.960773 - 91.843092
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 9.150978 1.254424 - 5.017696
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 1.183452 182.75677299999998 - 5848.216735999999
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 1.931728 283.875829 - 9084.026528
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 4.943807 12.994411 - 415.821152
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 16.567456 6.919445 - 27.67778
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 7.925028 7.833494 - 31.333976
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 11.404724 1.142653 - 4.570612
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 28.265948 0.6946986666666667 - 2.7787946666666667
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 24.221611 0.5377303333333333 - 2.1509213333333332
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 12.497374 0.22211166666666668 - 0.8884466666666667
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 21.303923 1.157875 - 4.6315
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 20.209841 0.5405866666666665 - 2.162346666666666
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 17.073282 0.12675333333333333 - 0.5070133333333333
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 10.592799 87.13177999999999 - 2788.2169599999997
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 5.71345 100.728905 - 3223.32496
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 5.810914 12.2403 - 391.6896
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 7.389816 1.344167612903226 - 43.013363612903234
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 13.675057 1.2736976451612902 - 40.75832464516129
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 7.03239 0.3106614193548387 - 9.941165419354839
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 4.981844 3.0638862258064514 - 98.04435922580645
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 10.070317 1.4112917741935482 - 45.16133677419354
Qwen3-4B-Instruct-2507 GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 8.015922 0.2074222580645161 - 6.637512258064516
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 17.024301 0.08906033333333332 - 0.3562413333333333
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1024 15.810341 0.092818 - 0.742544
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2048 13.337368 0.10240906666666666 - 1.6385450666666665
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 3072 12.53338 0.11044921739130434 - 2.650781217391304
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 11.425286 0.12383183870967741 - 3.9626188387096772
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 512 17.254415 0.091904 - 0.367616
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 1024 15.960709 0.10345585714285714 - 0.8276468571428571
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 2048 15.015873 0.10949759999999999 - 1.7519615999999998
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 3072 14.223588 0.1146168695652174 - 2.750804869565217
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 4096 13.424662 0.1365795483870968 - 4.370545548387097
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® X2 Elite 4096 42.6465 0.045206 - 1.446592
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Snapdragon® X Elite 4096 22.612881 0.098676 - 3.157632
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-8275 4096 17.863496 0.09899 - 3.16768
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 512 17.254415 0.091904 - 0.367616
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 1024 15.960709 0.10345585714285714 - 0.8276468571428571
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 2048 15.015873 0.10949759999999999 - 1.7519615999999998
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 3072 14.223588 0.1146168695652174 - 2.750804869565217
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-9075 4096 19.836973 0.09678 - 3.09696
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-X7181 4096 22.612881 0.098676 - 3.157632
Qwen3-4B-Instruct-2507 GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 4096 13.424662 0.1365795483870968 - 4.370545548387097

License

  • The license for the original implementation of Qwen3-4B-Instruct-2507 can be found here.

References

Community

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
Downloads last month

-

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

Paper for qualcomm/Qwen3-4B-Instruct-2507