LLM EXPLORER 66,704 MODELS INDEXED

LightOnOCR 3 4B MTP Repacked by prithivMLmods

By prithivMLmods · 662 downloads
HF Repo ↗ Copy URL

LightOnOCR 3 4B MTP Repacked is an open-source language model by prithivMLmods. Features: 4b LLM, VRAM: 9.3GB, License: apache-2.0.

Base model:lightonai/lightonoc... Base model:quantized:lightonai...   Conversational   Document-understanding   En   Endpoints compatible   Eval-results   Forms   Gguf   Image-text-to-text   Llama-cpp   Mtp   Ocr   Pdf   Qwen3 5   Region:us   Safetensors   Sharded   Tables   Tensorflow   Vision   Vision-language   Visual-grounding
Model Card on HF 🤗: https://huggingface.co/prithivMLmods/LightOnOCR-3-4B-MTP-Repacked 

LightOnOCR 3 4B MTP Repacked Parameters and Internals

LLM NameLightOnOCR 3 4B MTP Repacked
Repository 🤗https://huggingface.co/prithivMLmods/LightOnOCR-3-4B-MTP-Repacked 
Base Model(s)  LightOnOCR 3 4B   lightonai/LightOnOCR-3-4B
Model Size4b
Required VRAM9.3 GB
Updated2026-11-05
MaintainerprithivMLmods
Model Typeqwen3_5
Model Files  4.3 GB: 1-of-3   4.4 GB: 2-of-3   0.6 GB: 3-of-3
Supported Languagesen
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.5.4
Tokenizer ClassTokenizersBackend
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to LightOnOCR 3 4B MTP Repacked

Best Alternatives
Context / RAM
Downloads
Likes
AgentHorse 4B256K / 8.4 GB9737
Qwen3.5 4B0K / 9.3 GB6575502801
NuExtract30K / 0.2 GB441570324
LightOnOCR 3 4B0K / 9.1 GB239571
Agents A1 4B0K / 9.1 GB17757872
FrogNano 4B 26090K / 9.3 GB733113
Qwen3.5 4B Base0K / 9.3 GB22781479
UserRepGen 4B0K / 9.1 GB100813
Intern Decision 4B0K / 0.1 GB33848
Tev1 4B Experimental0K / 9.3 GB153916
Note: green Score (e.g. "73.2") means that the model is better than prithivMLmods/LightOnOCR-3-4B-MTP-Repacked.