Instructions to use LibreYOLO/LibreMolmo28b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LibreYOLO/LibreMolmo28b with Transformers:

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

pipe = pipeline("image-text-to-text", model="LibreYOLO/LibreMolmo28b", 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)

# Load model directly
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("LibreYOLO/LibreMolmo28b", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LibreYOLO/LibreMolmo28b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LibreYOLO/LibreMolmo28b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LibreYOLO/LibreMolmo28b",
		"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/LibreYOLO/LibreMolmo28b

SGLang

How to use LibreYOLO/LibreMolmo28b 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 "LibreYOLO/LibreMolmo28b" \
    --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": "LibreYOLO/LibreMolmo28b",
		"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 "LibreYOLO/LibreMolmo28b" \
        --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": "LibreYOLO/LibreMolmo28b",
		"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 LibreYOLO/LibreMolmo28b with Docker Model Runner:
```
docker model run hf.co/LibreYOLO/LibreMolmo28b
```

LibreMolmo28b

Snapshot mirror of allenai/Molmo2-8B at commit e28fa28597e5ec5e0cca2201dd8ab33d48bc4a1b for LibreYOLO single-image pointing. Weights, tokenizer assets, configuration and remote model code are unchanged. LibreYOLO adds this card, LICENSE and NOTICE; the original card is preserved in README.upstream.md.

Usage

Requires a LibreYOLO build containing the Molmo2 family. Install its dedicated extra in a separate environment: pip install 'libreyolo[molmo2]'. The pinned remote code requires Transformers 4.57.1.

from libreyolo import LibreVLM
model = LibreVLM("molmo2-8b", names=["boat"])
result = model.predict("image.jpg")
print(result.points.xy)

Pointing is the default and only supported prediction task. Confidence is a synthetic 1.0. Detection, training, validation, export and point tracking are not supported. Larger variants require more memory; no peak-memory or accuracy benchmark is claimed here.

License and provenance

Released by the Allen Institute for AI (Ai2) under Apache-2.0. See LICENSE, NOTICE, and the original upstream card. The runtime loads remote model code from a pinned snapshot.