Hugging Face、エッジ向けに高速・高品質な視覚機能を持つ LFM2.5-VL-3B を公開

Back to Articles

LFM2.5-VL-3B is our most capable vision-language model you can run on your own hardware. It understands documents and screens alike, grounds objects, and can call tools. It answers directly instead of reasoning, so responses stay fast in real-time and on-device apps.

LFM2.5-VL-3B extends the vision-language capabilities of our previous releases with four major improvements:

• Screen/UI understanding: Strong understanding of digital screens across different devices.

• Grounding: Improved grounding and object detection with natural language queries.

• Multi-image input: Improved reasoning across multiple images.

• Function calling: Significantly stronger at function calling, in text-only and vision-text situations.

How we trained our most capable vision-language model

LFM2.5-VL-3B pairs a SigLIP2 400M NaFlex vision encoder with the same pre-trained backbone as our LFM2.5-2.6B text model. It is pre-trained on about 34T tokens, with 4x more vision data than before, drawn from curated and synthetic image-caption, OCR, grounding, and instruction-following sets. To support non-Latin scripts, we doubled the vocabulary to 128K by extending the tokenizer in place rather than retraining from scratch.

Post-training runs in two stages: First is supervised fine-tuning (SFT), with knowledge distillation from a larger teacher and Antidoom training. Second is multi-reward reinforcement learning (RL).

Benchmark results

We evaluated LFM2.5-VL-3B across both vision and text benchmarks.

The vision benchmarks cover multilingual visual comprehension, instruction following, visual math and scientific reasoning, document understanding, object detection, multi-image understanding, and screen understanding. LFM2.5-VL-3B leads its size class on real-world image tasks, while also reading digital content well, from documents and charts to on-screen UI elements.

*All values in the table are normalized to 0–100. Evaluation is done using vLLM 0.26.0 and each model’s recommended generation parameters when available. Non-reasoning mode is used everywhere, and models are prompted to directly answer without reasoning.

We also evaluated LFM2.5-VL-3B on text-only benchmarks for instruction following and tool use. Instruction following climbs across the board, and tool use improves sharply. On tool use, LFM2.5-VL-3B is on par with Gemma-4-E2B and Qwen3.5-2B.

*InternVL 3.5 models do not support function-calling.

These results demonstrate that LFM2.5-VL-3B is a strong, general-purpose vision-language model. It covers everyday tasks (captioning, visual question answering, document understanding) and is especially good at grounding objects, reading screens and documents, and calling tools.

Inference speed on CPU and GPU

LFM2.5-VL-3B ships with day-one support across the inference ecosystem, including llama.cpp, MLX, vLLM, SGLang, and ONNX.

On-device inference. LFM2.5-VL-3B decodes 228 tokens/s on an M5 Max and 116 tokens/s on a Ryzen AI Max+ 395, and fits in about 3 GB of memory. It even reaches 20 tokens/s on a Galaxy S26 Ultra, so you can run it fully on-device.

GPU inference. LFM2.5-VL-3B keeps latency consistently low and is the fastest on multi-frame inputs.

LFM2.5-VL-3B is also the fastest on output throughput out of all models we tested, reaching about 11K tokens per second at high concurrency. That is roughly 2× the larger 4B-class models and ahead of even the smaller 2B-class models, which adds up to nearly 1B output tokens per day on a single H100.

How to use LFM2.5-VL-3B

Reach for LFM2.5-VL-3B when you need on-device intelligence for high-volume workloads.

Install the latest version of transformers (compatible with transformers>=5.0.0):

%pip install -q torch torchvision accelerate "transformers>=5.10.1"

Then load and run the model:

import torch
from transformers.image_utils import load_image
from transformers import AutoModelForImageTextToText, AutoProcessor
from IPython.display import display

MODEL_ID = "LiquidAI/LFM2.5-VL-3B"

processor = AutoProcessor.from_pretrained(MODEL_ID)
model = AutoModelForImageTextToText.from_pretrained(
    MODEL_ID,
    device_map="auto",
    dtype="bfloat16",
)

img_url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
input_image = load_image(img_url)
display(input_image)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": input_image},
            {"type": "text", "text": "Describe this image in two concise sentences."},
        ],
    }
]

inputs = processor.apply_chat_template(
    messages,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt",
).to(model.device)

with torch.inference_mode():
    outputs = model.generate(
        **inputs,
        do_sample=True,
        temperature=0.2,
        top_k=50,
        repetition_penalty=1.0,
        max_new_tokens=256,
    )

output = processor.batch_decode(outputs[:, inputs["input_ids"].shape[1]:], skip_special_tokens=True)[0]
print(output)

Two cats are sleeping on a pink couch with two remote controls.

You can find more hands-on examples on how to use LFM2.5-VL3B for multi-image inputs, grounding, OCR, tool calling, and more in our documentation. Check out our release blog for video examples.

LFM2.5-VL-3B demo

Check out this browser demo of LFM2.5-VL-3B powering a vision-capable chat interface. It allows you to take or upload multiple images and let the model interact with them, including grounding, OCR, and tool use.

Get Started

LFM2.5-VL-3B is available on Hugging Face today.

With LFM2.5, we're delivering on our vision of AI that runs anywhere. These models are:

• Download: LFM2.5-VL-3B on Hugging Face.

• Try: run the WebGPU demo in your browser, no setup needed.

• Fine-tune: adapt LFM2.5-VL-3B to your task with our fine-tuning tutorials.

We can't wait to see what you build.

Citation

Please cite this article as:

Liquid AI, "LFM2.5-VL-3B: A Better and Faster Vision-Language Model for the Edge", Liquid AI Blog, Aug 2026.

Or use the BibTeX citation:

@article{liquidAI2026VL3B,
  author  = {Liquid AI},
  title   = {LFM2.5-VL-3B: A Better and Faster Vision-Language Model for the Edge},
  journal = {Liquid AI Blog},
  year    = {2026},
  note    = {www.liquid.ai/blog/lfm2-5-vl-3b},
}