Granite Docling 258M needs about 1.0 GB of memory at BF16 (a 0.3 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 0.3 GB (BF16) to 0.3 GB (BF16).
Granite Docling 258M – 0.164B llama architecture with vision support
Granite Docling 258M is a 0.164B llama model with vision capabilities. It is available in BF16 quantization and supports a 8192 token context window. The model is released under the Apache‑2.0 license.
Good at
- Vision‑enabled text generation
- Contextual document understanding
- Small‑scale inference
Limits
- Requires BF16 quant only
- Estimated 1 GB VRAM at 8 k context
- No other quantizations available
Best for
- Document summarization
- Image captioning
- Lightweight inference on modest GPUs
Who it’s for. Developers needing a compact vision‑enabled LLM for document processing or image‑text tasks. Suitable for edge deployments with limited VRAM.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| BF16 | 0.3 GB | 1.0 GB | Fits | Get file → | |
| BF16 | 0.3 GB | 1.0 GB | Fits | Get file → |
The only available quant is BF16, with a file size of 0.33 GB. At 8 k context it requires approximately 1 GB of VRAM. This is suitable for GPUs with at least 2 GB of VRAM for safe operation.
What GPU do you need for Granite Docling 258M?
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 BF16 (0.3 GB file, about 1.0 GB in use).
Which Granite Docling 258M quant should you choose?
BF16 (0.3 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Granite Docling 258M
How much VRAM does Granite Docling 258M need?
About 1.0 GB at BF16 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 Granite Docling 258M on an 8 GB GPU?
Yes, with BF16: 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 Granite Docling 258M GGUF?
It depends on the quant: BF16 is 0.3 GB, BF16 is 0.3 GB.
What context length does Granite Docling 258M support?
The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.2 GB at 8K.
Where can I download Granite Docling 258M GGUF?
The files are on Hugging Face in the ibm-granite/granite-docling-258M-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 Granite Docling 258M 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 Granite Docling 258M on a GPU with 4 GB VRAM?
Yes, BF16 quant requires about 1 GB at 8 k context, leaving headroom for other processes.
Does this model support vision tasks?
Yes, it has detected vision abilities and can process image inputs.
What license applies to this model?
The model is released under the Apache‑2.0 license.