LLM EXPLORER 56,604 MODELS INDEXED

Nanonets OCR2 3B 8bit by mlx-community

By mlx-community · 103 downloads
HF Repo ↗ Copy URL

Nanonets OCR2 3B 8bit is an open-source language model by mlx-community. Features: 3b LLM, VRAM: 4.6GB, Context: 125K, Quantized, Instruction-Based, LLM Explorer Score: 0.21.

  8-bit   8bit Base model:quantized:qwen/qwen... Base model:qwen/qwen2.5-vl-3b-...   Conversational   Endpoints compatible   Image-text-to-text   Image-to-text   Instruct   Mlx   Multilingual   Ocr   Pdf2markdown   Quantized   Qwen2 5 vl   Region:us   Safetensors   Vqa
Model Card on HF 🤗: https://huggingface.co/mlx-community/Nanonets-OCR2-3B-8bit 

Nanonets OCR2 3B 8bit Parameters and Internals

LLM NameNanonets OCR2 3B 8bit
Repository 🤗https://huggingface.co/mlx-community/Nanonets-OCR2-3B-8bit 
Base Model(s)  Qwen/Qwen2.5-VL-3B-Instruct   Qwen/Qwen2.5-VL-3B-Instruct
Model Size3b
Required VRAM4.6 GB
Updated2026-08-08
Maintainermlx-community
Model Typeqwen2_5_vl
Instruction-BasedYes
Model Files  4.6 GB
Quantization Type8bit
Model ArchitectureQwen2_5_VLForConditionalGeneration
Context Length128000
Model Max Length128000
Transformers Version4.57.0
Padding Token<|endoftext|>
Vocabulary Size151936
Errorsreplace

Best Alternatives to Nanonets OCR2 3B 8bit

Best Alternatives
Context / RAM
Downloads
Likes
Qwen2.5 VL 3B Instruct 4bit125K / 3.1 GB32355
Holo1 3B 4bit125K / 3.1 GB271
Holo1 3B 3bit125K / 2.7 GB211
Qwen2.5 VL 3B Instruct 8bit125K / 4.1 GB5287
Qwen2.5 VL 3B Instruct 3bit125K / 2.7 GB2212
Qwen2.5 VL 3B Instruct 6bit125K / 3.2 GB983
Nanonets OCR2 3B125K / 7.5 GB659329511
ScaleCUA 3B125K / 7.5 GB26113
Qwen2.5 VL 3B Instruct125K / 7.5 GB27106
Nanonets OCR S Bf16125K / 7.5 GB962
Note: green Score (e.g. "73.2") means that the model is better than mlx-community/Nanonets-OCR2-3B-8bit.