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

Libraries

How to use microsoft/MagenticBrain with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/MagenticBrain")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("microsoft/MagenticBrain")
model = AutoModelForCausalLM.from_pretrained("microsoft/MagenticBrain", 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 microsoft/MagenticBrain with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/MagenticBrain"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/MagenticBrain",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/microsoft/MagenticBrain

SGLang

How to use microsoft/MagenticBrain 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 "microsoft/MagenticBrain" \
    --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": "microsoft/MagenticBrain",
		"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 "microsoft/MagenticBrain" \
        --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": "microsoft/MagenticBrain",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/MagenticBrain with Docker Model Runner:
```
docker model run hf.co/microsoft/MagenticBrain
```

aagohary commited on 6 days ago

Commit

6f48155

verified ·

1 Parent(s): d23c536

Document RL stage on in-house terminal tasks in model card

Browse files

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -19,7 +19,7 @@ tags:
 MagenticBrain is a 14B-parameter orchestration model from **Microsoft Research AI Frontiers**. It plans multi-step tasks, calls declared tools, and coordinates sub-agents. It does not execute actions itself — every real-world side effect happens inside a host harness.
-The model is supervised fine-tuned from [Qwen/Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B) on agentic data: function-calling corpora, file-system trajectories, terminal tasks, sub-agent delegation traces, and reasoning data. It's co-designed with **MagenticLite**, our agentic application and harness, and that's the configuration it has been most thoroughly evaluated in.
 We're releasing weights only. Inference code, training recipes, and the execution harness are part of MagenticLite.
@@ -144,7 +144,7 @@ MagenticBrain runs in non-thinking mode — keep `enable_thinking=False` in the
 ### Approach
-Post-training is Supervised Fine-Tuning on a heterogeneous agentic data mix. Reinforcement learning is in active exploration but not part of this release. Thinking tokens are disabled by default (`enable_thinking=False`) to control verbosity and reduce looping on long trajectories.
 ### Data sources

 MagenticBrain is a 14B-parameter orchestration model from **Microsoft Research AI Frontiers**. It plans multi-step tasks, calls declared tools, and coordinates sub-agents. It does not execute actions itself — every real-world side effect happens inside a host harness.
+The model is supervised fine-tuned from [Qwen/Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B) on agentic data: function-calling corpora, file-system trajectories, terminal tasks, sub-agent delegation traces, and reasoning data. After supervised fine-tuning, it was further trained with reinforcement learning on in-house-constructed terminal tasks. It's co-designed with **MagenticLite**, our agentic application and harness, and that's the configuration it has been most thoroughly evaluated in.
 We're releasing weights only. Inference code, training recipes, and the execution harness are part of MagenticLite.
 ### Approach
+Post-training has two stages. First, Supervised Fine-Tuning on a heterogeneous agentic data mix. Second, a reinforcement learning stage on in-house-constructed terminal tasks, further specializing the model for multi-step CLI execution. Thinking tokens are disabled by default (`enable_thinking=False`) to control verbosity and reduce looping on long trajectories.
 ### Data sources