Nuqta Arabic OCR 0.8B needs about 1.5 GB of memory at Q4_K_M (a 0.6 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 0.6 GB (Q4_K_M) to 0.9 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 1.8 GB.
Nuqta Arabic OCR 0.8B – lightweight Arabic OCR for phones
Nuqta Arabic OCR 0.8B is an open‑weights model with 0.81B parameters. It converts printed Arabic pages, photos and PDFs into clean text or Markdown with HTML tables, preserving tashkeel. The model runs fully offline on CPU or GPU.
Good at
- Arabic OCR with high accuracy
- Preserves tashkeel and tables
- Runs on phones and laptops
Limits
- No vision support beyond single image OCR
- Requires about 1.6 GB RAM
- License is Apache‑2.0
Best for
- Mobile OCR apps
- Document digitization on laptops
- Offline Arabic text extraction
Who it’s for. Developers building Arabic OCR solutions for mobile or desktop environments. Researchers needing a lightweight, open‑source Arabic OCR model that preserves diacritics and table structure.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 0.6 GB | 1.5 GB | Fits | Get file → | |
| Q8_0 | 0.9 GB | 1.8 GB | Fits | Get file → |
For 8 GB VRAM, Q4_K_M (0.56 GB file, ~1.5 GB RAM at 8 k context) is suitable. If more VRAM is available, Q8_0 (0.87 GB file, ~1.8 GB RAM) offers higher precision. Both are approximate memory estimates.
What GPU do you need for Nuqta Arabic OCR 0.8B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is Q8_0 (0.9 GB file, about 1.8 GB in use).
Which Nuqta Arabic OCR 0.8B quant should you choose?
Q4_K_M is the usual starting point for Nuqta Arabic OCR 0.8B: it balances file size and quality well. Here it is a 0.6 GB file.
The Q8_0 file is 0.9 GB, about 1.6 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Nuqta Arabic OCR 0.8B
How much VRAM does Nuqta Arabic OCR 0.8B need?
About 1.5 GB at Q4_K_M with an 8K context window. That is the 0.6 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Nuqta Arabic OCR 0.8B on an 8 GB GPU?
Yes, with Q8_0: the 0.9 GB file needs about 1.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Nuqta Arabic OCR 0.8B GGUF?
It depends on the quant: Q4_K_M is 0.6 GB, Q8_0 is 0.9 GB.
What context length does Nuqta Arabic OCR 0.8B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.8 GB at 32K.
Where can I download Nuqta Arabic OCR 0.8B GGUF?
The files are on Hugging Face in the SalahALHaismawi/Nuqta-Arabic-OCR-0.8B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Nuqta Arabic OCR 0.8B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Nuqta Arabic OCR on 8 GB VRAM?
Yes, the Q4_K_M quant requires about 1.5 GB RAM at 8 k context, well within 8 GB VRAM limits.
Does the model support images larger than 8 k context?
The trained context is 262,144 tokens; image size is limited by the multimodal implementation in llama.cpp.
Is the model licensed for commercial use?
It is released under Apache‑2.0, allowing commercial deployment with attribution.