You can deploy this model and start earning money today!

qualcomm/
Qwen3.5-2B

Name: Qwen3.5-2B
Availability: InStock

Views

No views yet

Qwen3.5-2B: Optimized for Qualcomm Devices

Qwen3.5 is the latest multilingual language model series from Alibaba Cloud with improved reasoning and instruction-following capabilities over Qwen3.

This is based on the implementation of Qwen3.5-2B 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 Qwen3.5-2B on-device

Follow the GenieX quickstart to install GenieX and deploy the model on a target 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
GENIEX_LLAMACPP	q4_0	Universal		Download

For more device-specific assets and performance metrics, visit Qwen3.5-2B 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 Qwen3.5-2B on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

Model architecture: Dense Transformer with Grouped Query Attention (GQA) and SwiGLU activation.
Response Rate: Rate of response generation after the first response token.
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.

Performance Summary

Model	Runtime	Precision	Chipset	Context Length	Response Rate (tokens per second)	Time To First Token (range, seconds)
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	512	39.677816	1.6498689999999998 - 6.599475999999999
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	512	29.470299	0.801365 - 3.20546
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	512	26.584835	0.539421 - 2.157684
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	4096	33.805635	22.612256000000002 - 723.5921920000001
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	4096	20.606342	12.676426 - 405.645632
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Gen 5 Mobile	4096	25.202407	4.972359 - 159.115488
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	512	38.95598	0.5104920000000001 - 2.0419680000000002
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	512	38.599606	0.50898375 - 2.035935
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	512	28.489984	0.20170275000000001 - 0.8068110000000001
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	4096	31.19537	0.72233153125 - 23.114609
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	4096	30.734418	0.761061 - 24.353952
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® 8 Elite Mobile	4096	25.497292	0.2283448125 - 7.307034
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	512	54.837315	0.589646 - 2.358584
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	512	40.60213	0.450651 - 1.802604
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	512	24.38805	0.510589 - 2.042356
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	4096	49.333135	6.54229 - 209.35328
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	4096	38.436846	3.856001 - 123.392032
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X2 Elite	4096	23.272165	4.054354 - 129.739328
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	512	34.175611	1.1716659999999999 - 4.6866639999999995
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	512	30.039833	1.020958 - 4.083832
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	512	18.767467	1.195416 - 4.781664
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	4096	37.674018	14.140977999999999 - 452.51129599999996
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	4096	28.592337	8.871029 - 283.872928
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Snapdragon® X Elite	4096	18.761124	9.478714 - 303.318848
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	512	13.022424	7.268928 - 29.075712
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	512	3.860738	9.915331 - 39.661324
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	512	8.823348	1.948441 - 7.793764
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	4096	8.782793	62.464046 - 1998.849472
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	4096	3.651685	86.556271 - 2769.800672
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-8275	4096	8.477796	14.966916999999999 - 478.94134399999996
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	512	29.52874	3.016486 - 12.065944
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	512	12.519398	3.534936 - 14.139744
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	512	8.013093	1.7871579999999998 - 7.148631999999999
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	512	34.175611	1.1716659999999999 - 4.6866639999999995
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	512	30.039833	1.020958 - 4.083832
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	512	18.767467	1.195416 - 4.781664
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	512	38.95598	0.5104920000000001 - 2.0419680000000002
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	512	38.599606	0.50898375 - 2.035935
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	512	28.489984	0.20170275000000001 - 0.8068110000000001
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	4096	24.800357	27.320434000000002 - 874.2538880000001
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	4096	11.3528	31.04587 - 993.46784
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-9075	4096	7.870393	13.517209000000001 - 432.55068800000004
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	4096	37.674018	14.140977999999999 - 452.51129599999996
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	4096	28.592337	8.871029 - 283.872928
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ IQ-X7181	4096	18.761124	9.478714 - 303.318848
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	4096	31.19537	0.72233153125 - 23.114609
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	4096	30.734418	0.761061 - 24.353952
Qwen3.5-2B	GENIEX_LLAMACPP	q4_0	Qualcomm® Dragonwing™ Q-8750	4096	25.497292	0.2283448125 - 7.307034

License

The license for the original implementation of Qwen3.5-2B can be found here.

References

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.

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

qualcomm/Qwen3.5-2B