Instructions to use LiquidAI/LFM3-VL-2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LiquidAI/LFM3-VL-2B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="LiquidAI/LFM3-VL-2B", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("LiquidAI/LFM3-VL-2B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LiquidAI/LFM3-VL-2B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LiquidAI/LFM3-VL-2B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LiquidAI/LFM3-VL-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/LiquidAI/LFM3-VL-2B

SGLang

How to use LiquidAI/LFM3-VL-2B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "LiquidAI/LFM3-VL-2B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LiquidAI/LFM3-VL-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "LiquidAI/LFM3-VL-2B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LiquidAI/LFM3-VL-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use LiquidAI/LFM3-VL-2B with Docker Model Runner:
```
docker model run hf.co/LiquidAI/LFM3-VL-2B
```

SaylorTwift HF Staff commited on 5 days ago

Commit

6d6c04a

verified ·

1 Parent(s): da10385

Add evaluation results

Browse files

# Add evaluation results for LiquidAI/LFM3-VL-2B

## Summary

This PR adds the benchmark results published on the model card for `LiquidAI/LFM3-VL-2B` to `.eval_results/`, following the [Hugging Face Hub evaluation-results specification](https://huggingface.co/docs/hub/eval-results).

## Benchmarks Added

- [MMMU-Pro (standard, 10 options)](https://huggingface.co/datasets/MMMU/MMMU_Pro?eval_result=LiquidAI/LFM3-VL-2B&leaderboard_task_id=mmmu_pro_standard_10_options) — 45.7
- [ScreenSpot-Pro](https://huggingface.co/datasets/likaixin/ScreenSpot-Pro?eval_result=LiquidAI/LFM3-VL-2B&leaderboard_task_id=overall) — 55.5

## Benchmarks Skipped (Not Registered on Hub)

These benchmarks are on the model card but have no registered `eval.yaml` on the Hub:

- The other 60 of the card's 62 vision benchmarks (MME, MMStar, DocVQA, ChartQA, MathVista, RefCOCO, ScreenSpot-v2, MVBench, ...)

They can be added once their authors register an `eval.yaml` on the Hub.

## Mapping notes

- Video-MME (62.8 without subtitles, 69.0 with): the card uses Video-MME v1, but the Hub registers Video-MME v2.
- Scores come from the reasoning-mode column; the no-thinking configuration is not added.

## Source

- Model card: https://huggingface.co/LiquidAI/LFM3-VL-2B
- Blog: https://www.liquid.ai/blog/lfm3

## Files Added

- `.eval_results/LFM3-VL-2B.yaml`

## Verification

The scores come from the benchmark tables on the model card, as measured by Liquid AI. No verified token is included, because these were not run via HF Jobs with inspect-ai.

---

**To upload:**

```bash
hf upload LiquidAI/LFM3-VL-2B .eval_results .eval_results --create-pr --commit-message "Add evaluation results"
```

Files changed (1) hide show

.eval_results/LFM3-VL-2B.yaml +16 -0

.eval_results/LFM3-VL-2B.yaml ADDED Viewed

	@@ -0,0 +1,16 @@

+- dataset:
+    id: MMMU/MMMU_Pro
+    task_id: mmmu_pro_standard_10_options
+  value: 45.7
+  date: "2026-10-27"
+  source:
+    url: https://huggingface.co/LiquidAI/LFM3-VL-2B
+    name: "LFM3-VL-2B model card"
+- dataset:
+    id: likaixin/ScreenSpot-Pro
+    task_id: overall
+  value: 55.5
+  date: "2026-10-27"
+  source:
+    url: https://huggingface.co/LiquidAI/LFM3-VL-2B
+    name: "LFM3-VL-2B model card"