NuExtract3 needs about 4.4 GB of memory at Q4_K_M (a 2.8 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 6 quants, from 2.0 GB (Q2_K) to 4.6 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.4 GB.
NuExtract3 – 4B vision‑language model for structured document extraction
NuExtract3 is a 4.3B parameter model that handles text and image inputs for document understanding. It performs structured extraction, image‑to‑Markdown conversion, and reasoning across multilingual documents. The model supports multimodal inputs and template generation.
Good at
- Structured extraction from invoices and forms
- Image‑to‑Markdown conversion
- Reasoning over long document contexts
Limits
- No vision support beyond images
- Requires large VRAM for full context
- Apache‑2.0 license limits commercial use without attribution
Best for
- OCR pipelines for receipts
- RAG preprocessing for contracts
- Automated form data extraction
Who it’s for. Developers building document processing workflows who need both text and image understanding. The model is suitable for teams that require structured JSON output and Markdown conversion from scanned documents.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 2.0 GB | 3.6 GB | Fits | Get file → | |
| Q3_K_M | 2.3 GB | 3.9 GB | Fits | Get file → | |
| Q4_K_M | 2.8 GB | 4.4 GB | Fits | Get file → | |
| Q5_K_M | 3.2 GB | 4.8 GB | Fits | Get file → | |
| Q6_K | 3.6 GB | 5.2 GB | Fits | Get file → | |
| Q8_0 | 4.6 GB | 6.2 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_M (2.78 GB file, ~4.4 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (3.16 GB file, ~4.8 GB memory) offers better quality. Higher quants like Q6_K or Q8_0 need >16 GB VRAM.
What GPU do you need for NuExtract3?
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 Q3_K_M (2.3 GB file, about 3.9 GB in use).
- 6 GB VRAM: the best quant that fits is Q6_K (3.6 GB file, about 5.2 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (4.6 GB file, about 6.2 GB in use).
Which NuExtract3 quant should you choose?
Q4_K_M is the usual starting point for NuExtract3: it balances file size and quality well. Here it is a 2.8 GB file.
The Q8_0 file is 4.6 GB, about 1.7 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.
If memory is tight, the smallest file is Q2_K at 2.0 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about NuExtract3
How much VRAM does NuExtract3 need?
About 4.4 GB at Q4_K_M with an 8K context window. That is the 2.8 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 NuExtract3 on an 8 GB GPU?
Yes, with Q8_0: the 4.6 GB file needs about 6.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of NuExtract3 GGUF?
It depends on the quant: Q2_K is 2.0 GB, Q3_K_M is 2.3 GB, Q4_K_M is 2.8 GB, Q5_K_M is 3.2 GB, Q6_K is 3.6 GB, Q8_0 is 4.6 GB.
What context length does NuExtract3 support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.4 GB at 32K.
Where can I download NuExtract3 GGUF?
The files are on Hugging Face in the numind/NuExtract3-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 NuExtract3 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 NuExtract3 on 8 GB VRAM?
Yes, using the Q4_K_M quantized model which requires about 4.4 GB of VRAM at 8 k context.
Does NuExtract3 support vision tasks?
Yes, it accepts images as part of multimodal input for document understanding.