unsloth/gemma-4-E2B-it

Hugging Face
VISIONConcurrency Cost:1Model Size:5.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Mar 31, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

The unsloth/gemma-4-E2B-it is a 5.1 billion parameter instruction-tuned multimodal model developed by Google DeepMind, part of the Gemma 4 family. It supports text, image, and audio inputs with text output, featuring a 32K token context window. This model is optimized for on-device deployment, excelling in reasoning, coding, and agentic workflows, with native function-calling support.

Loading preview...

Overview

unsloth/gemma-4-E2B-it is an instruction-tuned variant from the Gemma 4 family, developed by Google DeepMind. This model is a multimodal powerhouse, capable of processing text, image, and audio inputs (with audio natively supported on this E2B variant) to generate text outputs. It features a 5.1 billion total parameter count (2.3B effective) and a 128K token context window, making it suitable for efficient local execution on devices like laptops and mobile phones.

Key Capabilities

  • Multimodal Understanding: Processes text, images (with variable aspect ratio and resolution), and audio (ASR, speech-to-translated-text).
  • Reasoning: Includes a built-in reasoning mode for step-by-step thinking.
  • Coding & Agentic Capabilities: Enhanced performance in coding benchmarks and native function-calling support for autonomous agents.
  • Long Context: Supports a 128K token context window.
  • Multilingual: Pre-trained on 140+ languages with out-of-the-box support for 35+ languages.

What Makes It Different?

This E2B model is specifically optimized for on-device deployment due to its efficient architecture, including Per-Layer Embeddings (PLE) for parameter efficiency. It offers a strong balance of multimodal capabilities and performance in a smaller footprint, making it ideal for applications requiring local execution. The model also introduces native system prompt support for more structured conversations and configurable thinking modes.

Should You Use This?

This model is an excellent choice for developers building applications that require:

  • On-device multimodal AI: Ideal for mobile or edge deployments needing text, image, and audio processing.
  • Reasoning and agentic workflows: Its enhanced reasoning and function-calling capabilities are beneficial for complex tasks.
  • Coding assistance: Strong performance in code generation, completion, and correction.
  • Long context understanding: Handles prompts up to 128K tokens, useful for detailed analysis or summarization.

Consider its limitations regarding factual accuracy and common sense, as with most LLMs, and implement appropriate safety safeguards.

Finetunes
17 models