UNUnlimited-OCR
Unlimited-OCR
Model Overview
Unlimited-OCR is a 3.3B-parameter Mixture-of-Experts image text to text model developed by baidu. Built on the UnlimitedOCRForCausalLM architecture. Supports multilingual. Released on 2026-06-19.
📊 Quick Specs
| Specification | Value |
|---|---|
| Parameters | 3.3B |
| Architecture | UnlimitedOCRForCausalLM |
| Task | image text to text |
| Modality | text, image |
| License | MIT |
| Framework | transformers |
| MoE | Yes |
| Languages | multilingual |
✨ Key Features
- 3.3B parameters with sparse MoE architecture for efficient inference
- Built on UnlimitedOCRForCausalLM architecture (transformers)
- Primary task: image text to text (text, image modality)
- Language support: multilingual
- Open-weights under MIT license — self-hostable and fine-tunable
- High adoption: 2,500,391 downloads on Hugging Face
📈 Community Adoption
- 3,029 likes on Hugging Face
- 2,500,391 downloads on Hugging Face
🔗 Resources
- Hugging Face Hub: Unlimited-OCR on Hugging Face
- Paper: arXiv
📜 License & Access
MIT — Open-weights model available for download, fine-tuning, and self-hosted deployment.
Key Features
3.3B parameters with sparse MoE architecture for efficient inference
Built on UnlimitedOCRForCausalLM architecture (transformers)
Primary task: image text to text (text, image modality)
Language support: multilingual
Open-weights under MIT license — self-hostable and fine-tunable
High adoption: 2,500,391 downloads on Hugging Face
You might also want to compare
Verified Sources
Tags
Model Specs
Parameters
3.3B
Context Window
unknown
License
MIT
Deployment
Resources & Links
Curator Notes
Auto-ingested from HuggingFace Trending on 2026-07-25. Architecture: UnlimitedOCRForCausalLM. Params: 3.3B. License: MIT.
Compare Specs
Compare parameters, context windows, modalities, and benchmark scores of this model side-by-side with others.
Compare Model