Gemma-4-E2B 8da4w

Mobile CPUvivo X300SME2

A compact instruction-tuned autoregressive language model for text generation and multiple-choice scoring, provided as a text-only quantized variant with dynamic 8-bit activations, grouped 4-bit linear weights, and packed 4-bit embeddings; multimodal encoders are not included.View on Hugging Face

Tokens / sec

15.66 tok/s

Time to First Token

709 ms

Latency

8883.5 ms

Peak Memory

2008.3 MB

Get Started with this Model

Learning path

Run optimized text-to-text models from the Arm AI Portal on Arm-powered Android devices

Run optimized text-generation and embedding models from the Arm AI Portal using supported Android runtimes and model adapters.

Learning path

Build a Sentiment-Aware Voice Assistant with On-Device LLMs

Build an end-to-end, on-device voice assistant that understands both speech and emotion using Whisper, HuBERT, ONNX Runtime, and a local LLM with llama.cpp on Arm.

Learning path

Build a customer support chatbot on Android with Llama and ExecuTorch

Learn how to build a customer support chatbot for Android using Llama 3.2, ExecuTorch, and KleidiAI to run on-device inference on Arm platforms.

Learning path

Build an Android chat application with ONNX Runtime API

Learn how to build ONNX Runtime and the generate() API for Android to run a Phi-3 model on Arm-based smartphones.

Model Evaluation

Base model: gemma-4-E2BTested on: vivo X300 (arm64, 8-core)

Evaluation Conditions

Text Generation · vivo X300 · INT4 · MMLU-Pro

Metric	Resultsvs baseline	Change
End to end latency p50	8883.50ms33382.50ms	−24499.00 ms
End to end latency p90

About this version

Original Model:: gemma-4-E2B by Google — Repository
Optimization/conversion:: Arm-Optimized version for execution on Arm-based platforms.
Converted/optimized by:: Arm
License:: The Original Model and the Optimized Model are subject to Apache-2.0.

This repository contains a converted or optimized version of the Original Model (the “Optimized Model”). The Original Model has been converted or optimized as described above for execution on Arm-based platforms.

No retraining or fine-tuning of the Original Model was performed as part of the conversion or optimization. The conversion or optimization was not intended to change the Original Model’s behavior or intended use.

Original Model and Documentation

For information about the Original Model, including its development, training data, intended uses, limitations and other relevant information, please refer to the Original Model repository. Information in that repository was provided by the original developer or other third parties and, unless expressly stated otherwise, has not been independently verified by Arm.

Licenses and Third-Party Terms

Use of the Original Model and the Optimized Model is subject to the applicable licenses, usage restrictions and other terms identified above and in the relevant repositories. Publication of the Optimized Model does not grant any rights beyond those provided under the applicable license terms.

You are responsible for reviewing those terms and ensuring that your use of the Original Model and the Optimized Model is permitted.

Configuration	Baseline	Arm Optimized
Dataset	MMLU-Pro	MMLU-Pro
Sample count	1	1
Batch size	1	1

Gemma-4-E2B 8da4w

Get Started with this Model

Run optimized text-to-text models from the Arm AI Portal on Arm-powered Android devices

Build a Sentiment-Aware Voice Assistant with On-Device LLMs

Build a customer support chatbot on Android with Llama and ExecuTorch

Build an Android chat application with ONNX Runtime API

Model Evaluation

Gemma-4-E2B 8da4w

Get Started with this Model

Run optimized text-to-text models from the Arm AI Portal on Arm-powered Android devices

Build a Sentiment-Aware Voice Assistant with On-Device LLMs

Build a customer support chatbot on Android with Llama and ExecuTorch

Build an Android chat application with ONNX Runtime API

Model Evaluation

Configuration

Comparability

Purpose of this Release