Build and run AI on-device.

The complete edge AI workflow for developers.

Get started Documentation

Terminal

user@computer:~$

hub.embedl.com

Fine-tuning results

end-to-end toolbox

store data and artifacts

Supported hardware platforms

No fluff. Just edge AI tools that work.

Explore the full workflow.

On-device benchmarks

Compare models and hardware to find the best combination for on-device performance.

Python library

Fine-tune your model on your data. Compile, quantize and verify performance on your device.

Experiment tracking

Store data and artifacts on the web. Analyze and visualize your results and KPI:s.

It all starts with the best model.

Browse the largest on-device AI benchmark suite.

compare open-source models and edge hardware

hub.embedl.com

Fine-tuning results

Train it. Optimize it. Run it.

Deploy your model for any edge device with the Hub Python library.

pick and choose components from our toolbox

Adapt your model.

Use our training recipes for easy fine-tuning on your own data.

user@computer:~$

Smaller faster models.

Optimize your model for lower latency and memory usage.

user@computer:~$ embedl-hub quantize \
-m my_model.onnx \
--data /path/to/calibration_dataset/ \
--num-samples 1000

Target every chip.

Compile your model for execution on CPU, GPU, NPU or other AI accelerators on your target devices.

user@computer:~$ embedl-hub compile \
-m my_quantized_model.onnx \
--device "Samsung Galaxy S25" \
--runtime tflite

Run your own benchmarks.

Measure latency and memory usage of your model on a real edge device in the cloud.

user@computer:~$ embedl-hub profile \
-m my_model.tflite \
--device "Samsung Galaxy S25"

All your results in one place.

Analyze and visualize your experiments on the web.

store data and artifacts

get an overview of all KPI:s

hub.embedl.com

Fine-tuning results

hub.embedl.com

Fine-tuning results

Start building for free.

The complete edge AI workflow that works.

JOIN THE

"I used to spend more time in dependency hell than in VS Code, that's why we built this."

Hans Salomonsson
CEO, Embedl

"You might never go back to cloud inference again! Since this is the first BETA, we had to use our own quotes here."

Andreas Ask
Embedded Engineer

EDGE AI

MOVEMENT

Frequently asked questions

This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 780681, No 957197 and No 190109922.