Run LightOnOCR-3 0.8B and Other Open Models on a Free Colab T4
Melvin Vivas · X post · 2026-10-21 · Open on X
Topics: LLM Fundamentals, Fine-tuning & Model Customization, AI Dev Tools & Productivity · Level: intermediate
Summary
Melvin Vivas added a notebook for the LightOnOCR-3 0.8B vision/OCR model to his public GitHub repo of Colab notebooks. The repo also has notebooks for newer open models that fit on a Google Colab T4 GPU instance, so you can try local-model inference and fine-tuning without paying for hardware.
Key points
- LightOnOCR-3 is a small (0.8B-parameter) vision model for OCR, small enough to run on a free Colab T4 GPU.
- The donvito/notebooks repo collects notebooks for fine-tuning and running local models, all made to run in Google Colab.
- The repo also has notebooks for other new open models that fit on a T4 GPU instance.
- A T4 GPU is a good limit to aim for: models under about 1B parameters, or quantized larger ones, fit comfortably.
Resources mentioned
- donvito/notebooks · repo · github.com · free
The creator's notebooks for fine-tuning and running local models, which you can run in Google Colab, including a GLiNER2.5-Decide intent classification example.
Also in: Try Liquid AI's open d1-3B decision model in Colab (Melvin Vivas on X · notes), Try EmbeddingGemma 2 multimodal embeddings in Colab (Melvin Vivas on X · notes), Colab Notebook for Liquid AI's d1-3B Decision Model (Melvin Vivas on X · notes), Fine-tune and Run Local Models for Free on Google Colab's T4 GPU (Melvin Vivas on X · notes) and 10 more - LightOnOCR-3 0.8B · tool · huggingface.co · free
A small open vision model for OCR (text extraction from images and documents). - Google Colab · tool · colab.research.google.com · free · recommended by both Bashiri Smith & Melvin Vivas
Free GPU notebooks.
Also in: Try Liquid AI's open d1-3B decision model in Colab (Melvin Vivas on X · notes), Google Colab notebooks for new open models, built by a coding agent (Melvin Vivas on X · notes), Liquid AI's LFM2.5-24B-A2B: a local MoE model for agent tasks (Melvin Vivas on X · notes), A100 80GB GPUs in Google Colab via Google AI Pro plans (Melvin Vivas on X · notes) and 27 more
Try this
- Open the LightOnOCR-3 notebook in the donvito/notebooks repo and run it on a free Colab T4 GPU.
- Build an OCR pipeline with LightOnOCR-3 on Colab that pulls text out of scanned documents or images.