moondream2 needs about 5.0 GB of memory at F16 (a 2.8 GB file) with an 8K context window, so it fits on a 6 GB GPU.
Moondream 2 – a 1.4 B parameter vision‑language model in GGUF format
Moondream 2 is a small vision‑language model built on the phi2 architecture. It supports image understanding and grounded reasoning with a 2048‑token context window. The model is released under the Apache‑2.0 license and is available as a single F16 GGUF file.
Good at
- Grounded visual reasoning
- Object detection and UI element localization
- Fast text generation with superword tokenizer
Limits
- Requires 5 GB VRAM at 8 k context (F16)
- No multilingual support beyond English
- No vision‑language tasks beyond those trained
Best for
- Image captioning and description
- UI automation and screen analysis
- Document OCR and table extraction
Who it’s for. Developers building lightweight vision‑language applications on edge devices or with limited GPU memory. The model is suitable for projects that need quick visual reasoning without large compute resources.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| F16 | 2.8 GB | 3.7 GB | Fits | Get file → |
The only available quant is F16, 2.84 GB in size. It needs approximately 5 GB VRAM for 8 k context. If your GPU has less than 5 GB, the model will not run with this quant. No other quants are provided, so memory budget must match this requirement.
What GPU do you need for moondream2?
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: no quant fits at an 8K context.
- 6 GB VRAM: the best quant that fits is F16 (2.8 GB file, about 5.0 GB in use).
Which moondream2 quant should you choose?
F16 (2.8 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 moondream2
How much VRAM does moondream2 need?
About 5.0 GB at F16 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 moondream2 on an 8 GB GPU?
Yes, with F16: the 2.8 GB file needs about 5.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of moondream2 GGUF?
It depends on the quant: F16 is 2.8 GB.
What context length does moondream2 support?
The GGUF header lists a trained context length of 2,048 tokens (2K). The KV cache grows with the context you use: about 1.6 GB at 8K and 0.4 GB at 2K.
Where can I download moondream2 GGUF?
The files are on Hugging Face in the ggml-org/moondream2-20250414-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 moondream2 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 Moondream 2 on 8 GB VRAM?
Yes, the F16 quant requires about 5 GB VRAM at 8 k context, leaving headroom for other processes.
Does Moondream 2 support multilingual text?
No, the model is trained only on English language data and does not support other languages.
Is there a vision‑language model with a smaller memory footprint?
No other quants are available for Moondream 2; the F16 file is the only format provided.