LLM EXPLORER 57,918 MODELS INDEXED

Gemma 3 4B It by unsloth

By unsloth · 62743 downloads
HF Repo ↗ Copy URL

Gemma 3 4B It is an open-source language model by unsloth. Features: 4b LLM, VRAM: 8.6GB, License: gemma, LLM Explorer Score: 0.24.

  Arxiv:1705.03551   Arxiv:1810.12440   Arxiv:1903.00161   Arxiv:1904.09728   Arxiv:1905.07830   Arxiv:1905.10044   Arxiv:1907.10641   Arxiv:1908.02660   Arxiv:1910.11856   Arxiv:1911.01547   Arxiv:1911.11641   Arxiv:2009.03300   Arxiv:2103.03874   Arxiv:2104.12756   Arxiv:2106.03193   Arxiv:2107.03374   Arxiv:2108.07732   Arxiv:2110.14168   Arxiv:2203.10244   Arxiv:2210.03057   Arxiv:2304.06364   Arxiv:2311.12022   Arxiv:2311.16502   Arxiv:2312.11805   Arxiv:2404.12390   Arxiv:2404.16816   Arxiv:2502.12404   Arxiv:2502.21228 Base model:finetune:google/gem... Base model:google/gemma-3-4b-i...   Conversational   Deploy:azure   Endpoints compatible   Gemma3   Image-text-to-text   Region:us   Safetensors   Sharded   Tensorflow   Unsloth
Model Card on HF 🤗: https://huggingface.co/unsloth/gemma-3-4b-it 

Gemma 3 4B It Parameters and Internals

LLM NameGemma 3 4B It
Repository 🤗https://huggingface.co/unsloth/gemma-3-4b-it 
Base Model(s)  Gemma 3 4B It   google/gemma-3-4b-it
Model Size4b
Required VRAM8.6 GB
Updated2026-08-10
Maintainerunsloth
Model Typegemma3
Model Files  5.0 GB: 1-of-2   3.6 GB: 2-of-2
Model ArchitectureGemma3ForConditionalGeneration
Licensegemma
Model Max Length131072
Transformers Version4.51.3
Tokenizer ClassGemmaTokenizer
Padding Token<pad>
Torch Data Typebfloat16

Best Alternatives to Gemma 3 4B It

Best Alternatives
Context / RAM
Downloads
Likes
Medgemma 4B It Abliterated128K / 6.3 GB136
Gemma 3 Glitter 4B128K / 8.6 GB3347
Fallen Gemma3 4B V1128K / 8.6 GB1318
Gemma 3 4B It0K / 8.6 GB18552621345
Translategemma 4B It0K / 8.6 GB38998793
Nemotron 3 Content Safety0K / 8.6 GB170618
...on Content Safety Reasoning 4B0K / 10 GB869029
Medgemma 1.5 4B It0K / 8.6 GB29369
Gemma 3 4B Pt0K / 8.6 GB72499159
...Gemma 3 4B It Datav3 Only Mask0K / 8.6 GB910
Note: green Score (e.g. "73.2") means that the model is better than unsloth/gemma-3-4b-it.