Instructions to use MidTool/Arctic-MidTool-RL-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MidTool/Arctic-MidTool-RL-4B with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MidTool/Arctic-MidTool-RL-4B")
model = AutoModelForCausalLM.from_pretrained("MidTool/Arctic-MidTool-RL-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 MidTool/Arctic-MidTool-RL-4B with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/MidTool/Arctic-MidTool-RL-4B

SGLang

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

Docker Model Runner
How to use MidTool/Arctic-MidTool-RL-4B with Docker Model Runner:
```
docker model run hf.co/MidTool/Arctic-MidTool-RL-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.

Arctic-MidTool-RL-4B

The final 4B agent from the MidTool recipe: Qwen3-4B-Base → mid-training on MidTool-Mix → tool-use SFT → agentic RL. Start here if you want a model to run, rather than one to train further (Arctic-MidTool-MT-4B is the mid-training checkpoint).

Results

4B model	BFCLv3 Overall	τ²-Bench Pass@1	MCP-Universe Score
Qwen3-4B (official)	24.27	11.87	16.05
Qwen3-4B-Base + SFT + RL	39.51	13.04	17.65
Arctic-MidTool-RL-4B	54.18	19.96	19.73

Largest gains are on multi-turn BFCL (27.63 vs 19.00) and τ²-Bench retail/airline — the settings that need long-horizon execution and recovery.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "MidTool/Arctic-MidTool-RL-4B"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype="auto", device_map="auto")

messages = [{"role": "user", "content": "What's the weather in Seattle?"}]
inputs = tok.apply_chat_template(messages, tools=my_tools,
                                 add_generation_prompt=True, return_tensors="pt")

Tool calls are emitted in the Qwen3 <tool_call> format; pass your schemas via the tools= argument of the chat template. Thinking mode is disabled in our evaluation setup.

Details

See our paper for the full data, training, and evaluation details.

@article{jiang2026midtool,
  title  = {MidTool: Mid-training Data Synthesis for Agentic Tool Use},
  author = {Jiang, Fengqing and Wang, Yite and Liu, Boyi and Wang, Zhaoyang and
            Xu, Canwen and Yao, Zhewei and Poovendran, Radha and He, Yuxiong},
  year   = {2026}
}

Downloads last month: -

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for MidTool/Arctic-MidTool-RL-4B

Base model

Qwen/Qwen3-4B-Base

Finetuned

MidTool/Arctic-MidTool-MT-4B

Finetuned

(1)

this model