Unlimited-OCR-GGUF
GGUF quantizations of baidu/Unlimited-OCR with its vision projector, for llama.cpp and Ollama.
What it outputs: Unlimited-OCR is a document-layout OCR model. It returns each text region with its type and a bounding box, e.g. title [34, 88, 370, 172]INVOICE 4471, and marks pictures as image [x1, y1, x2, y2] regions instead of describing them. Use a general vision model for image descriptions.
| File | What |
|---|---|
Unlimited-OCR-Q4_K_M.gguf |
Language model, Q4_K_M (recommended) |
Unlimited-OCR-Q8_0.gguf |
Language model, Q8_0 |
mmproj-Unlimited-OCR-F16.gguf |
Vision projector (f16), required for images |
Ollama
ollama run richardyoung/unlimited-ocr
Tags: Q4_K_M (default, latest), Q8_0. Image input works in the CLI (drag in a file path) and via the API images field.
llama.cpp
llama-mtmd-cli -m Unlimited-OCR-Q4_K_M.gguf --mmproj mmproj-Unlimited-OCR-F16.gguf --image page.png -p "Transcribe all the text in this image."
Verified
Tested before upload through Ollama with a synthetic invoice and a shapes image (Q4_K_M shown):
- Invoice transcription: 'title [34, 88, 370, 172]INVOICE 4471 / text [34, 249, 317, 309]Date: 2026-09-28 / text [34, 362, 418, 432]Widgets x3 $42.00 / text [34, 487, 384, 573]TOTAL: $86.20'
- Shapes: 'image [90, 301, 382, 752] image [513, 301, 816, 752]'
Converted with llama.cpp convert_hf_to_gguf.py (text and --mmproj), quantized with llama-quantize.