Instructions to use aisquared/bolt-vl-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use aisquared/bolt-vl-4b with Transformers:

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

pipe = pipeline("image-text-to-text", model="aisquared/bolt-vl-4b")
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 AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("aisquared/bolt-vl-4b")
model = AutoModelForMultimodalLM.from_pretrained("aisquared/bolt-vl-4b")
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?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use aisquared/bolt-vl-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "aisquared/bolt-vl-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "aisquared/bolt-vl-4b",
		"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/aisquared/bolt-vl-4b

SGLang

How to use aisquared/bolt-vl-4b 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 "aisquared/bolt-vl-4b" \
    --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": "aisquared/bolt-vl-4b",
		"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 "aisquared/bolt-vl-4b" \
        --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": "aisquared/bolt-vl-4b",
		"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"
						}
					}
				]
			}
		]
	}'

Unsloth Studio

How to use aisquared/bolt-vl-4b with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aisquared/bolt-vl-4b to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aisquared/bolt-vl-4b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for aisquared/bolt-vl-4b to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="aisquared/bolt-vl-4b",
    max_seq_length=2048,
)

Docker Model Runner
How to use aisquared/bolt-vl-4b with Docker Model Runner:
```
docker model run hf.co/aisquared/bolt-vl-4b
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Model Overview

bolt-vl-4b is a specialized multimodal Large Language Model (LLM) fine-tuned of Qwen 3.5. Designed by AISquared, this model excels in document analysis, image-to-markdown conversion, and general conversational assistance involving visual inputs.

Invoice parsing performance vs. model size

Key Capabilities

Document Conversion: Converts scanned invoices and text-heavy documents into structured Markdown, including tables and headings.
Visual Reasoning: Interprets visual elements (logos, signatures, stamps) without hallucinating external data.

System Prompt & Usage

When using the model, ensure you provide a clear system prompt to activate the Bolt Assistant persona and the Markdown conversion capability.

Recommended System Prompt

For general tasks:

You are the Bolt assistant. Your primary task is to analyze images of documents and convert their content into structured Markdown. You must adhere strictly to the content present in the image, ignoring any physical artifacts like hole punches or stamps unless they contain valid information. Do not output commentary; only return the clean Markdown.

For invoice parsing:

You are a document conversion assistant. Your task is to convert the provided document image into well-structured Markdown. You MUST:
    1. Extract ALL visible text content exactly as it appears.
    2. Preserve the document's structure — use Markdown tables for tabular data, headings for section titles, lists for itemized content, etc.
    3. For any non-text visual elements (logos, stamps, signatures, graphics), insert a brief italic description in brackets, e.g. *[Company logo]*. DO NOT include e.g. hole punches or any other commentary about the document. Remember, stick to just the content on the doc!
    4. Do NOT add any commentary, explanation, or preamble — output ONLY the Markdown representation of the document.

Benchmark Performance

OmniDocBench & OCRBenchV2

Summary of OmniDocBench and OCRBenchV2 Results. Normalized so that higher is better for all scores.

AISquared Invoice Benchmark

Percentage of entities (line items and invoice fields) that were extracted from invoices with 100% accuracy.

Model Limitations & Disclaimer

OCR Accuracy: While fine-tuned for text extraction, extremely low-resolution or handwritten images may result in transcription errors.
Context Window: The model was trained on a maximum sequence length of 16k tokens. For documents larger than this limit, the model may truncate the beginning or end of the input unless specifically handled via sliding windows (not default behavior).