Rapid42/gemma-4-E4B-it-MLX

Gemma 4 (~8B, E4B variant) — MLX format for Apple Silicon, instruction-tuned

Converted and optimized by Rapid42 — engineering tools for fast pipelines.

What This Is

This is Gemma 4 E4B (Google DeepMind's 4th generation Gemma, ~8B parameters in the E4B multimodal variant) converted to MLX format for native Apple Silicon inference. Instruction-tuned (-it) for chat and task-following.

Gemma 4 is Google's latest open model family — multimodal (text + image input), with strong performance on reasoning, coding, and multilingual tasks.

Parameters: ~8B (E4B = Efficient 4B-class, actual ~8B)
Modality: Text + Image input → Text output
Format: MLX (Apple Silicon native)
Base model: google/gemma-4-it
License: Apache 2.0

Hardware Requirements

Device	RAM	Experience
Any M-series Mac (16GB+)	~8GB	✅ Fast and smooth
M1 MacBook Air (8GB)	~8GB	⚠️ Tight — works but little headroom
M3 Max / Pro	~8GB	✅ Near-instant

A practical choice for multimodal tasks on any Apple Silicon machine.

Quick Start

pip install mlx-lm

Text chat:

from mlx_lm import load, generate

model, tokenizer = load("Rapid42/gemma-4-E4B-it-MLX")

messages = [{"role": "user", "content": "What are the key advantages of MLX over PyTorch for Apple Silicon?"}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_dict=False
)

response = generate(model, tokenizer, prompt=prompt, max_tokens=512, verbose=True)
print(response)

With image input (via MLX-VLM):

pip install mlx-vlm
python -m mlx_vlm.generate \
  --model Rapid42/gemma-4-E4B-it-MLX \
  --prompt "Describe what you see in this image." \
  --image /path/to/image.jpg