SmolVLM2-256M-Video-Instruct needs about 0.8 GB of memory at Q4_K_M (a 0.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 4 quants, from 0.1 GB (Q4_K_M) to 0.3 GB (F16).
SmolVLM2-256M-Video-Instruct: lightweight multimodal model for video, image, and text tasks
SmolVLM2-256M-Video-Instruct is a 0.163 B parameter multimodal model based on the Idefics3 architecture. It processes video, image, and text inputs to generate text outputs such as captions, visual question answering, and storytelling. The model is licensed under Apache 2.0 and is optimized for on‑device inference.
Good at
- Video and image analysis
- Text‑based queries on media
- Low‑resource inference
Limits
- No image or video generation
- Requires 8 k context for full performance
- Limited to English language
Best for
- Captioning short videos
- Answering visual questions
- Storytelling from media
Who it’s for. Developers needing a compact multimodal model for on‑device or edge inference. Suitable for applications that require video or image analysis with limited GPU memory.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 0.1 GB | 0.8 GB | Fits | Get file → | |
| Q8_0 | 0.2 GB | 0.9 GB | Fits | Get file → | |
| BF16 | 0.3 GB | 1.0 GB | Fits | Get file → | |
| F16 | 0.3 GB | 1.0 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_M (0.13 GB file, ~0.8 GB memory at 8 k context). For 12 GB VRAM, Q8_0 (0.18 GB file, ~0.9 GB memory) offers higher precision. BF16 and F16 use ~1 GB memory and are suitable for 16 GB VRAM setups.
What GPU do you need for SmolVLM2-256M-Video-Instruct?
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 (0.3 GB file, about 1.0 GB in use).
Which SmolVLM2-256M-Video-Instruct quant should you choose?
Q4_K_M is the usual starting point for SmolVLM2-256M-Video-Instruct: it balances file size and quality well. Here it is a 0.1 GB file.
The Q8_0 file is 0.2 GB, about 1.3 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.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about SmolVLM2-256M-Video-Instruct
How much VRAM does SmolVLM2-256M-Video-Instruct need?
About 0.8 GB at Q4_K_M with an 8K context window. That is the 0.1 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 SmolVLM2-256M-Video-Instruct on an 8 GB GPU?
Yes, with F16: 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 SmolVLM2-256M-Video-Instruct GGUF?
It depends on the quant: Q4_K_M is 0.1 GB, Q8_0 is 0.2 GB, BF16 is 0.3 GB, F16 is 0.3 GB.
What context length does SmolVLM2-256M-Video-Instruct 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 SmolVLM2-256M-Video-Instruct GGUF?
The files are on Hugging Face in the ggml-org/SmolVLM2-256M-Video-Instruct-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 SmolVLM2-256M-Video-Instruct 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 SmolVLM2-256M-Video-Instruct on 8 GB VRAM?
Yes, the Q4_K_M quantization requires about 0.8 GB of VRAM at 8 k context, which fits within 8 GB.
Does the model support image generation?
No, the model only performs analysis and text generation based on visual inputs.
What license applies to this model?
The model is released under the Apache 2.0 license.