Model HQ
DocumentationHardware optimized
Qualcomm Supported Models
A complete catalog of 95 AI models optimized for Qualcomm Snapdragon processors using QNN runtime.
Hover the Encyclopedia tab on the right, or tap any icon to learn what a model type does.
95
Optimized models
QNN
Runtime
CPU · GPU · NPU
Hardware targets
Why Qualcomm
Qualcomm optimization features
Snapdragon-optimized models deliver faster inference by leveraging the Qualcomm Neural Network (QNN) SDK across Snapdragon hardware.
Performance benefits
- Optimized for Qualcomm Snapdragon architectures
- Enhanced inference speed with QNN runtime
- Power-efficient execution on mobile and edge devices
- Hardware-specific optimizations for Hexagon DSP and Adreno GPU
Supported hardware
- Qualcomm Snapdragon processors
- Hexagon Digital Signal Processors (DSP)
- Adreno GPUs
- Qualcomm AI Engine and NPUs
Catalog
CPU Models
GGUF and tool models that run on the CPU alone — no GPU or NPU required.
Agentic
23 modelsslim-sentiment-tool1.1Bslim-extract-tool1.1Bslim-summary-tool1.1Bslim-boolean-tool1.1Bslim-tags-tool1.1Bslim-xsum-tool1.1Bslim-emotions-tool1.1Bslim-topics-tool1.1Bslim-sql-tool1.1Bslim-extract-qwen-1.5b1.5Bslim-ner-tool1.1Bslim-sa-ner-tool1.1Bslim-tags-3b-tool3Bslim-extract-tiny-tool1.1Bslim-ratings-tool1.1Bslim-intent-tool1.1Bslim-category-tool1.1Bslim-nli-tool1.1Bslim-q-gen-phi-3-tool3.8Bslim-summary-tiny-tool1.1Bslim-qa-gen-phi-3-tool3.8Bslim-qa-gen-tiny-tool1.1Bslim-summary-tiny1.1BCoding
1 modelsqwen2.5-7b-coder7BGeneral Chat
10 modelstiny-llama-chat1.1Bllama-2-7b-chat7Bqwen2.5-32b32Bbling-qwen-0.5b0.5Bbling-qwen-1.5b1.5Bopenhermes-2.5-mistral7Bdragon-llama-3.18Bzephyr-7b-beta7Bstarling-lm-7b-alpha7BminiCPM-V-2_64BInstruct
12 modelsqwen-2.5-14b-instruct14Bqwen2-7B-instruct7Bqwen-2-0.5b-instruct0.5Bqwen2-1.5-instruct1.5Bllama-3.1-instruct8Bllama-3.2-ib-instruct3Bllama-3.2-1b-instruct1.1Bllama-3-8b-instruct8Bmistral-7b-instruct-v0.37Bgemma-2-9b-instruct9Bgemma-2-27b-instruct27Bgemma-2b-it2BQuestion-answer
11 modelsdragon-yi-answer-tool6Bdragon-llama-answer-tool7Bdragon-mistral-answer-tool7Bbling-answer-tool1.1Bbling-tiny-llama1.1Bbling-phi-33.8Bbling-phi-3.53.8Bdragon-mistral-0.37Bdragon-yi-9b9Bdragon-qwen-7b7Bbling-stablelm-3b3BCatalog
GPU/CPU/NPU Models
OpenVINO (OV) and ONNX models that run on the available GPU, CPU, or NPU.
Agentic
15 modelsslim-extract-phi-33.8Bslim-boolean-phi-33.8Bslim-summary-phi-33.8Bslim-emotions1.1Bslim-topics1.1Bslim-sql1.1Bslim-sentiment1.1Bslim-extract-tiny1.1Bslim-intent1.1Bslim-tags1.1Bslim-ratings1.1Bslim-ner1.1Bslim-extract-qwen-nano0.5Bslim-sa-ner-phi-33.8Bslim-xsum-phi-33.8BGeneral Chat
3 modelsllama-2-chat7Bphi-33.8Bphi-3.53.8BInstruct
10 modelsqwen-2.5-14b-instruct14Bqwen2-7B-instruct7Bqwen-2-0.5b-instruct0.5Bqwen2-1.5-instruct1.5Bllama-3.2-3b-instruct-onnx3Bllama-3.1-instruct8Bllama-3.2-1b-instruct1.1Bllama-3-8b-instruct8Bmistral-7b-instruct-v0.37Bgemma-2b-it2BPrompt Safety
3 modelsprotectai-prompt-injection0.3Bunitary-unbiased-toxic-roberta0.1Bvalurank-distilroberta-bias0.1BQuestion-answer
3 modelsdragon-mistral-0.37Bdragon-yi-9b9Bdragon-qwen-7b7BRe-ranker
2 modelsjina-reranker-turbo0.6Bjina-reranker-tiny0.1BVision
1 modelsphi-3-vision4.2BCatalog
NPU Models
Models designed to run on Neural Processing Units for efficient, low-power inference.
Vision
1 modelsllama-3.2-3b-onnx-qnn3BNext stepsCheck system requirements
Getting started with Qualcomm models
- 01Ensure you have a Qualcomm Snapdragon processor with QNN support.
- 02Select models optimized for Qualcomm from the Models section.
- 03The system automatically applies Qualcomm optimizations when available.
- 04Monitor performance improvements in the system metrics.
Not sure what your hardware supports?
Check the system requirements to find the right models for your machine.
For Qualcomm-specific optimization questions, contact our technical support team at support@aibloks.com.
Reference