gemma-4-E2B-it-litert-lm

gemma-4-E2B-it-litert-lm

Gemma 4 E2B instruction-tuned model packaged as a .litertlm file for the LiteRT-LM framework, made for private on-device use on phones, desktops and edge devices.

Made by litert-community on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: gemma-4-E2B-it-litert-lm is the Gemma 4 E2B instruction-tuned model from Google, repackaged by litert-community. It comes as a .litertlm file for LiteRT-LM, a framework for running generative models on a device.

The card names Android, iOS, desktop, IoT and the web as places it can run. Running it on the device keeps your use private and works without an internet connection.

Who it suits: It suits app developers and edge engineers who want a small language model inside their own app or device. People who only want to chat can try it through Google's AI Edge Gallery app, which the card links.

Industry | Who uses it | How they use it | Example
Mobile apps | Android and iOS developers | Put a small model inside an app | An offline assistant feature
IoT and edge | Embedded engineers | Run a model on a device without a network link | A local helper on a gadget
Desktop software | Developers | Add on-device text generation to a program | A writing aid that keeps data local
Privacy-minded teams | Product teams | Keep prompts on the user's device | A private note summarizer

Skip it if you want a large hosted model and have no need to run on a device.

Where to use it:

• The LiteRT-LM framework, which reads .litertlm files
• Android, iOS, desktop, IoT and web targets, per the card
• Google AI Edge Gallery, the showcase app built on the same stack
• Windows, macOS or Linux for the file download, with Python

Use case | Who | What they get back
On-device chat | App developers | Text replies with no network needed
Function calling | Agent builders | The model can call functions through the LiteRT-LM APIs
Low-memory runs | Edge engineers | Text-only use with weights as small as 0.8 GB

How to set it up: You need a few things in place first.

• Python and pip, for the download command
• The LiteRT-LM framework for your target platform
• Memory for the model, as low as 0.8 GB for text-only use according to the card

1. Install the Hugging Face hub package with pip.
2. Download the repository with the hf command.
3. Load the .litertlm file in LiteRT-LM. The model card and the LiteRT-LM documentation hold the run details.

pip install -U huggingface_hub
hf download litert-community/gemma-4-E2B-it-litert-lm

How to use it: You use the file through LiteRT-LM, which handles prompt templates, the key-value cache and function calling. The card also points to the Google AI Edge Gallery app and to a LiteRT-LM command line tool for trying the model.

Example: You put the file in a mobile app built on LiteRT-LM and ask the model to summarize a short note. The text stays on the phone, and the card says the model works without an internet connection.

For newcomers and experts: The easiest first try is the Google AI Edge Gallery app, which the card links for Android and iOS. Developers can then move on to the file itself.

The card describes a mobile quantization scheme that mixes 2-bit, 4-bit and 8-bit weights. For text-only use the weights can fit in as little as 0.8 GB, while memory mapping serves the 1.12 GB of embedding parameters. The vision and audio models load on demand to save more memory. LiteRT supplies XNNPack for the CPU and ML Drift for the GPU.

Download the .litertlm file
Load it in LiteRT-LM
Send a prompt on the device
Get a reply without a network

Good to know:

• License: Apache 2.0
• Cost to run: none named
• Data it reaches: it downloads files from Hugging Face, and once loaded the model runs on the device

On Hugging Face, 13 November 2026

• Library: litert-lm
• Downloads in the last 30 days: 1,418,552
• Likes: 545
• License: apache-2.0
• Based on: google/gemma-4-E2B-it
• Fine-tunes: 7
• Tags: litert-lm

Links

• Model: https://huggingface.co/litert-community/gemma-4-E2B-it-litert-lm
• Files: https://huggingface.co/litert-community/gemma-4-E2B-it-litert-lm/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download litert-community/gemma-4-E2B-it-litert-lm

Then follow the usage in its model card. Nothing in this kit runs by itself.