Model reference · open weights

Real-ESRGAN-x4plus

Available as managed deployment Image qualcomm Image edit 1 variants 1k dl/mo

Real-ESRGAN-x4plus is an open-weight image model from qualcomm. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.

Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.

What it is

Released byqualcomm
TypeImage models
TaskImage edit
Runs withpytorch
Released2024-02-25
Popularity1k downloads / month
LicenceOpen weights

About

What Real-ESRGAN-x4plus is

Real-ESRGAN is a machine learning model that upscales an image with minimal loss in quality. The implementation is a derivative of the Real-ESRGAN-x4plus architecture, a larger and more powerful version compared to the Real-ESRGAN-general-x4v3 architecture.

This is based on the implementation of Real-ESRGAN-x4plus 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.

Read the full model card

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.

RuntimePrecisionChipsetSDK VersionsDownload
ONNXfloatUniversalQAIRT 2.45, ONNX Runtime 1.27.1Download
ONNXw8a8UniversalQAIRT 2.45, ONNX Runtime 1.27.1Download
QNN_DLCfloatUniversalQAIRT 2.45Download
QNN_DLCw8a8UniversalQAIRT 2.45Download
TFLITEfloatUniversalQAIRT 2.45Download
TFLITEw8a8UniversalQAIRT 2.45Download

For more device-specific assets and performance metrics, visit Real-ESRGAN-x4plus 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 Real-ESRGAN-x4plus on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.super_resolution

Model Stats:

  • Input resolution: 128x128
  • Model checkpoint: RealESRGAN_x4plus
  • Model size (float): 63.9 MB
  • Model size (w8a8): 16.7 MB
  • Number of parameters: 16.7M

Performance Summary

ModelRuntimePrecisionChipsetInference Time (ms)Peak Memory Range (MB)Primary Compute Unit
Real-ESRGAN-x4plusONNXfloatSnapdragon® X2 Elite34.535 ms8 - 8 MBNPU
Real-ESRGAN-x4plusONNXfloatSnapdragon® X Elite65.478 ms37 - 37 MBNPU
Real-ESRGAN-x4plusONNXfloatSnapdragon® 8 Gen 3 Mobile50.898 ms0 - 743 MBNPU
Real-ESRGAN-x4plusONNXfloatSnapdragon® 8 Gen 1 Mobile142.871 ms4 - 739 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® Dragonwing™ IQ-8275127.753 ms6 - 10 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® Dragonwing™ QCS8550 (Proxy)69.852 ms0 - 44 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® QCS8450142.871 ms4 - 739 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® Dragonwing™ IQ-9075105.978 ms6 - 9 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® Dragonwing™ IQ-X718165.478 ms37 - 37 MBNPU
Real-ESRGAN-x4plusONNXfloatQualcomm® Dragonwing™ Q-875038.121 ms7 - 361 MBNPU
Real-ESRGAN-x4plusONNXfloatSnapdragon® 8 Elite Mobile38.121 ms7 - 361 MBNPU
Real-ESRGAN-x4plusONNXfloatSnapdragon® 8 Elite Gen 5 Mobile26.619 ms5 - 361 MBNPU
Real-ESRGAN-x4plusONNXw8a8Snapdragon® X2 Elite10.228 ms3 - 3 MBNPU
Real-ESRGAN-x4plusONNXw8a8Snapdragon® X Elite25.116 ms22 - 22 MBNPU
Real-ESRGAN-x4plusONNXw8a8Snapdragon® 8 Gen 3 Mobile16.475 ms2 - 759 MBNPU
Real-ESRGAN-x4plusONNXw8a8Snapdragon® 8 Gen 1 Mobile38.123 ms2 - 770 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ QCS6490146.398 ms2 - 4 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ IQ-827522.822 ms2 - 5 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ QCS8550 (Proxy)23.854 ms0 - 29 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® QCS845038.123 ms2 - 770 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ IQ-907522.533 ms2 - 4 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ IQ-X718125.116 ms22 - 22 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ Q-6690468.801 ms1 - 506 MBNPU
Real-ESRGAN-x4plusONNXw8a8Qualcomm® Dragonwing™ Q-779038.237

From the published model card. Full card on the HuggingFace links in the sidebar.

How it works

How image models work

Text promptwhat to makeText encoderunderstands itDiffusion stepsdenoise to pixelsImagePNG / JPEGA diffusion model starts from noise and denoises it, guided by your prompt, into a finished image.

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys real-esrgan-x4plus for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (real-esrgan-x4plus below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/images/generations \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"real-esrgan-x4plus","prompt":"a red bicycle","size":"1024x1024"}'

Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms