Welcome to LEAP!
Liquid Edge AI Platform (LEAP) is an integrated toolchain that enables developers to effectively specialize and deploy small language models to their use case on their hardware.
LEAP is designed to help any app developer – regardless of their familiarity with AI – to unlock the best quality and fastest performance for their edge AI task. Our end-to-end ecosystem of tools simplifies the entire edge AI app development process: from model discovery, through customization, to deployment.
1. Find the model you need
Find the ideal model based on your specific use-case requirements and constraints using our model search tool.
Browse and download our online catalog of pre-trained, pre-quantized model bundles for download.
2. Vibe-check the model on device with 0 setup
Download and vibe-check LEAP models on your own device with our mobile app for iOS and Android.
Vibe-check LFM2 models in a virtual sandbox to test their response quality via API.
3. Deploy the model to your target device
Deploy LEAP models - either from the Model Library or the Model Bundling Service - in your mobile app as easily as calling a cloud API.
Leverage llama.cpp for blazing fast inference on laptops - particularly on AMD Ryzen™ chip technology.
4. Customize the model to boost performance
Base models not good enough? Optionally customize a LEAP base model even further to your use case using our suite of tools to help simplify the finetuning process.
Bundle your finetuned checkpoints on-demand for usage with the Edge SDK (not required for LEAP base models).