nanoLLaVA needs about 2.6 GB of memory at F16 (a 1.2 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 3.2 GB.
NanoLLaVA: 0.62B Qwen2 model with vision capability
NanoLLaVA is a 0.62 billion‑parameter Qwen2 model that supports vision tasks. It is distributed as a single GGUF file and licensed under Apache‑2.0. The model was trained with a 32 768‑token context window.
Good at
- Vision‑enabled LLM tasks
- Small footprint for edge use
- Supports 32 768‑token context
Limits
- No mixture of experts
- Only one quant available (F16)
- Requires ~2.6 GB VRAM at 8 k context
Best for
- Image captioning on low‑resource devices
- Quick vision‑LLM prototyping
- Embedded vision applications
Who it’s for. Developers needing a lightweight vision‑enabled LLM for edge or mobile deployments. The model is suitable for projects that can run with limited GPU memory and do not require large‑scale inference.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| F16 | 1.2 GB | 2.6 GB | Fits | Get file → |
The only available quant is F16, with a file size of 1.25 GB. It requires approximately 2.6 GB VRAM at an 8 k context length. This is suitable for systems with at least 4 GB of VRAM.
What GPU do you need for nanoLLaVA?
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 F16 (1.2 GB file, about 2.6 GB in use).
Which nanoLLaVA quant should you choose?
F16 (1.2 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 nanoLLaVA
How much VRAM does nanoLLaVA need?
About 2.6 GB at F16 with an 8K context window. That is the 1.2 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 nanoLLaVA on an 8 GB GPU?
Yes, with F16: the 1.2 GB file needs about 2.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of nanoLLaVA GGUF?
It depends on the quant: F16 is 1.2 GB.
What context length does nanoLLaVA support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download nanoLLaVA GGUF?
The files are on Hugging Face in the abetlen/nanollava-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 nanoLLaVA 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 NanoLLaVA on 4 GB VRAM?
Yes, the F16 quant requires about 2.6 GB VRAM at 8 k context, so a 4 GB GPU can handle it with some headroom.
Does NanoLLaVA support text‑only tasks?
Yes, it can perform standard LLM tasks, but its primary strength is vision‑enabled processing.
Is there a smaller quant available?
No, only the F16 quant is provided in the current GGUF release.