LFM2.5 VL 450M needs about 1.3 GB of memory at Q4_K_M (a 0.2 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 8 quants, from 0.2 GB (Q4_0) to 1.4 GB (F32). Longer context needs more: at 32K tokens the KV cache alone adds about 2.1 GB.
LFM2.5‑VL‑450M: 0.354B vision‑language model with 32,768‑token context
LFM2.5‑VL‑450M is a 0.354 B parameter vision‑language model built on LFM2.5‑350M backbone and SigLIP2 NaFlex vision encoder. It supports 32,768‑token context, multilingual vision understanding, bounding‑box prediction, and function calling for text input.
Good at
- Multilingual vision understanding
- Bounding‑box prediction
- Function calling support
Limits
- No knowledge‑intensive tasks
- Not suited for fine‑grained OCR
- No vision‑only inference without text
Best for
- Image captioning
- Grounded visual question answering
- Object detection in images
Who it’s for. Developers building on‑device vision‑language applications that require a balance of speed and memory efficiency. Suitable for edge AI deployments where VRAM is limited.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_0 | 0.2 GB | 1.3 GB | Fits | Get file → | |
| Q4_K_M | 0.2 GB | 1.3 GB | Fits | Get file → | |
| Q5_K_M | 0.3 GB | 1.3 GB | Fits | Get file → | |
| Q6_K | 0.3 GB | 1.3 GB | Fits | Get file → | |
| Q8_0 | 0.4 GB | 1.4 GB | Fits | Get file → | |
| BF16 | 0.7 GB | 1.7 GB | Fits | Get file → | |
| F16 | 0.7 GB | 1.7 GB | Fits | Get file → | |
| F32 | 1.4 GB | 2.5 GB | Fits | Get file → |
For 8 GB VRAM, Q4_0, Q4_K_M, or Q5_K_M are suitable, each around 0.22–0.26 GB with ~1.3 GB memory at 8k context. For 16 GB VRAM, Q6_K or Q8_0 offer higher precision while staying under 2 GB memory. Larger FP16 or BF16 variants need more than 1.5 GB memory and are best for high‑end GPUs.
What GPU do you need for LFM2.5 VL 450M?
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 F32 (1.4 GB file, about 2.5 GB in use).
Which LFM2.5 VL 450M quant should you choose?
Q4_K_M is the usual starting point for LFM2.5 VL 450M: it balances file size and quality well. Here it is a 0.2 GB file.
The Q8_0 file is 0.4 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 Q4_0 at 0.2 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about LFM2.5 VL 450M
How much VRAM does LFM2.5 VL 450M need?
About 1.3 GB at Q4_K_M with an 8K context window. That is the 0.2 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 LFM2.5 VL 450M on an 8 GB GPU?
Yes, with F32: the 1.4 GB file needs about 2.5 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of LFM2.5 VL 450M GGUF?
It depends on the quant: Q4_0 is 0.2 GB, Q4_K_M is 0.2 GB, Q5_K_M is 0.3 GB, Q6_K is 0.3 GB, Q8_0 is 0.4 GB, BF16 is 0.7 GB and 2 more.
What context length does LFM2.5 VL 450M support?
The GGUF header lists a trained context length of 128,000 tokens (125K). The KV cache grows with the context you use: about 0.5 GB at 8K and 2.1 GB at 32K.
Where can I download LFM2.5 VL 450M GGUF?
The files are on Hugging Face in the LiquidAI/LFM2.5-VL-450M-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 LFM2.5 VL 450M use?
Hugging Face lists the license as other. 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 LFM2.5‑VL‑450M on 8 GB VRAM?
Yes, Q4_0, Q4_K_M, or Q5_K_M quantizations fit within 8 GB VRAM and use about 1.3 GB memory at 8k context.
Does the model support image‑only inference?
No, the model requires text input; it is not designed for vision‑only inference.
What languages does the model understand?
English, Arabic, Chinese, French, German, Japanese, Korean, Portuguese, and Spanish.