PaddleOCR VL manga needs about 1.0 GB of memory at Q4_K_M (a 0.3 GB file) with an 8K context window, so it fits on a 4 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 0.6 GB.
Japanese manga OCR model in GGUF format
PaddleOCR VL manga is a fine‑tuned OCR model for Japanese manga text. It is based on PaddleOCR‑VL and trained on Manga109‑s and synthetic data. The model is available in GGUF format for use with llama.cpp and similar tools.
Good at
- High accuracy on manga speech bubbles
- Handles stylized fonts and full‑width characters
- Supports Japanese text recognition
Limits
- Requires vision input only
- No vision support for other modalities
- License is Apache‑2.0
Best for
- OCR of manga pages
- Extracting text from speech bubbles
- Japanese text extraction in images
Who it’s for. Developers building OCR solutions for manga or Japanese text. The model can be integrated with libraries that support PaddleOCR‑VL or llama.cpp.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 0.3 GB | 1.0 GB | Fits | Get file → |
The Q4_K_M quantization file is 0.3 GB and needs about 1 GB of VRAM at 8 k context. For 8 GB VRAM, this quant is suitable. No other quantizations are provided.
What GPU do you need for PaddleOCR VL manga?
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 Q4_K_M (0.3 GB file, about 1.0 GB in use).
Which PaddleOCR VL manga quant should you choose?
Q4_K_M is the usual starting point for PaddleOCR VL manga: it balances file size and quality well. Here it is a 0.3 GB file.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about PaddleOCR VL manga
How much VRAM does PaddleOCR VL manga need?
About 1.0 GB at Q4_K_M with an 8K context window. That is the 0.3 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 PaddleOCR VL manga on an 8 GB GPU?
Yes, with Q4_K_M: the 0.3 GB file needs about 1.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of PaddleOCR VL manga GGUF?
It depends on the quant: Q4_K_M is 0.3 GB.
What context length does PaddleOCR VL manga support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.6 GB at 32K.
Where can I download PaddleOCR VL manga GGUF?
The files are on Hugging Face in the LunarOilRig/PaddleOCR-VL-manga-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 PaddleOCR VL manga 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 this model on 8 GB VRAM?
Yes, the Q4_K_M quantization requires approximately 1 GB of VRAM at 8 k context, which fits within an 8 GB budget.
Does the model support other languages?
The model is specialized for Japanese manga text and does not include other language support.
What input does the model expect?
It expects a crop of a single text region from a manga image, not a full page.