LLM EXPLORER 59,516 MODELS INDEXED

Gemma 4 E4B It MLX 8bit by unsloth

By unsloth · 760 downloads
HF Repo ↗ Copy URL

Gemma 4 E4B It MLX 8bit is an open-source language model by unsloth. Features: 2.6b LLM, VRAM: 9GB, License: apache-2.0, Quantized, LLM Explorer Score: 0.39, ELO: 1318.

  8-bit   8bit Base model:google/gemma-4-e4b-... Base model:quantized:google/ge...   Conversational   Gemma   Gemma4   Google   Image-text-to-text   Quantized   Region:us   Safetensors   Sharded   Tensorflow   Unsloth
Model Card on HF 🤗: https://huggingface.co/unsloth/gemma-4-E4B-it-MLX-8bit 

Gemma 4 E4B It MLX 8bit Benchmarks

LMSys ELO 1318 vs 1272 (so35) 3.6%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

Gemma 4 E4B It MLX 8bit Parameters and Internals

LLM NameGemma 4 E4B It MLX 8bit
Repository 🤗https://huggingface.co/unsloth/gemma-4-E4B-it-MLX-8bit 
Base Model(s)  Gemma 4 E4B It   google/gemma-4-E4B-it
Model Size2.6b
Required VRAM9 GB
Updated2026-08-10
Maintainerunsloth
Model Typegemma4
Model Files  5.0 GB: 1-of-2   4.0 GB: 2-of-2
Quantization Type8bit
Model ArchitectureGemma4ForConditionalGeneration
Licenseapache-2.0
Transformers Version5.5.0.dev0
Tokenizer ClassGemmaTokenizer
Padding Token<pad>

Best Alternatives to Gemma 4 E4B It MLX 8bit

Best Alternatives
Context / RAM
Downloads
Likes
Gemma 4 E4b It 8bit0K / 8.9 GB205326
Gemma 4 E4b It Mxfp80K / 8.7 GB3719
Note: green Score (e.g. "73.2") means that the model is better than unsloth/gemma-4-E4B-it-MLX-8bit.