Instructions to use FractalAIResearch/Fathom-Synthesizer-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FractalAIResearch/Fathom-Synthesizer-4B with Transformers:

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

pipe = pipeline("text-generation", model="FractalAIResearch/Fathom-Synthesizer-4B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FractalAIResearch/Fathom-Synthesizer-4B")
model = AutoModelForCausalLM.from_pretrained("FractalAIResearch/Fathom-Synthesizer-4B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use FractalAIResearch/Fathom-Synthesizer-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FractalAIResearch/Fathom-Synthesizer-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": "FractalAIResearch/Fathom-Synthesizer-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FractalAIResearch/Fathom-Synthesizer-4B

SGLang

How to use FractalAIResearch/Fathom-Synthesizer-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 "FractalAIResearch/Fathom-Synthesizer-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": "FractalAIResearch/Fathom-Synthesizer-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "FractalAIResearch/Fathom-Synthesizer-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": "FractalAIResearch/Fathom-Synthesizer-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FractalAIResearch/Fathom-Synthesizer-4B with Docker Model Runner:
```
docker model run hf.co/FractalAIResearch/Fathom-Synthesizer-4B
```

Add pipeline tag, additional tags, and key links to model card

by nielsr HF Staff - opened Oct 9, 2025

base: refs/heads/main

←

from: refs/pr/3

Discussion Files changed

+15

-7

Files changed (1) hide show

README.md +15 -7

README.md CHANGED Viewed

@@ -1,11 +1,23 @@
 ---
-license: mit
-library_name: transformers
 base_model:
 - Qwen/Qwen3-4B
 ---
 # Fathom-DeepResearch: Unlocking Long Horizon Information Retrieval And Synthesis For SLMs
 <div align="center">
@@ -334,8 +346,4 @@ We would like to acknowledge the following works for enabling our project:
   note={Hugging Face},
   year={2025}
 }
-```
----

 ---
 base_model:
 - Qwen/Qwen3-4B
+library_name: transformers
+license: mit
+pipeline_tag: text-generation
+tags:
+- text-generation
+- agent
+- web-search
+- tool-use
+- information-retrieval
+- report-generation
+- qwen
+- RAG
 ---
 # Fathom-DeepResearch: Unlocking Long Horizon Information Retrieval And Synthesis For SLMs
+[[Paper]](https://huggingface.co/papers/2509.24107) - [[Code]](https://github.com/FractalAIResearchLabs/Fathom-DeepResearch) - [[Project Page]](https://github.com/FractalAIResearchLabs/Fathom-DeepResearch)
 <div align="center">
   note={Hugging Face},
   year={2025}
 }
+```