TinyDoc-VLM
256MApache 2.0CPU-Runnable Document Vision Language Model
TinyDoc-VLM 256M is a cpu-runnable document vision language model. Operates across Parameters: 256M · Architecture: SigLIP + SmolLM2 · Tasks: OCR · Layout · VQA · Table Extraction · Hardware: CPU-runnable · No GPU required · Format: HuggingFace Transformers · ONNX · License: Apache 2.0. License: Apache 2.0.
A 256M-parameter vision-language model for document understanding. Handles OCR, layout analysis, table extraction, and visual QA — all without a GPU. Built for regulated industries that process high volumes of documents privately.
- ✓Runs entirely on CPU — no GPU costs
- ✓Processes invoices, receipts, forms, and contracts
- ✓698+ downloads on HuggingFace
- ✓Designed for air-gapped and regulated environments
Get started in minutes
pip install transformers torch pillowfrom transformers import AutoProcessor, AutoModelForVision2Seq
from PIL import Image
processor = AutoProcessor.from_pretrained("eulogik/TinyDoc-VLM-256M")
model = AutoModelForVision2Seq.from_pretrained("eulogik/TinyDoc-VLM-256M")
image = Image.open("document.png")
inputs = processor(images=image, text="Extract all text", return_tensors="pt")
output = model.generate(**inputs, max_new_tokens=512)
print(processor.decode(output[0], skip_special_tokens=True))Cite this Model & Architecture
If you utilize TinyDoc-VLM in your academic research, benchmarks, or enterprise deployments, please cite our open-weight publication:
@misc{eulogik2026tinydocvlm256m,
title = {TinyDoc-VLM 256M: CPU-Runnable Document Vision Language Model},
author = {Kishore, Gautam and Eulogik Systems Engineering},
year = {2026},
publisher = {Eulogik},
howpublished = {\url{https://eulogik.com/models/tinydoc-vlm-256m}}
}Need help deploying TinyDoc-VLM?
We offer consulting, custom fine-tuning, and on-premise deployment for all our models.