Instructions to use mindlab-research/Macaron-V1-Preview-744B-Merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mindlab-research/Macaron-V1-Preview-744B-Merged with Transformers:

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

pipe = pipeline("text-generation", model="mindlab-research/Macaron-V1-Preview-744B-Merged")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("mindlab-research/Macaron-V1-Preview-744B-Merged")
model = AutoModelForCausalLM.from_pretrained("mindlab-research/Macaron-V1-Preview-744B-Merged", 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 mindlab-research/Macaron-V1-Preview-744B-Merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mindlab-research/Macaron-V1-Preview-744B-Merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mindlab-research/Macaron-V1-Preview-744B-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mindlab-research/Macaron-V1-Preview-744B-Merged

SGLang

How to use mindlab-research/Macaron-V1-Preview-744B-Merged 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 "mindlab-research/Macaron-V1-Preview-744B-Merged" \
    --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": "mindlab-research/Macaron-V1-Preview-744B-Merged",
		"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 "mindlab-research/Macaron-V1-Preview-744B-Merged" \
        --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": "mindlab-research/Macaron-V1-Preview-744B-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use mindlab-research/Macaron-V1-Preview-744B-Merged with Docker Model Runner:
```
docker model run hf.co/mindlab-research/Macaron-V1-Preview-744B-Merged
```

Macaron-V1-Preview-744B-Merged

Macaron-V1-Preview-744B-Merged is a full-checkpoint merged variant of Macaron-V1-Preview from MindLab Research, post-trained from GLM-5.1 with MinT.

Release blog · Routed 749B MoL release

Model Overview

Field	Value
Model name	Macaron-V1-Preview-744B-Merged
Organization	MindLab Research
Base model	GLM-5.1
Architecture	Merged full checkpoint
Parameter footprint	744B-class
Post-training system	MinT
Primary domain	Personal agents, tool-use agents, Generative UI
Release type	Preview
Checkpoint format	Full checkpoint at repository root
Context length	202,752 tokens, from `config.json` / `tokenizer_config.json`
Precision	bfloat16, from `config.json`
License	MIT

Loading

Install minimal loading dependencies:

pip install -U transformers accelerate safetensors

Example:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

repo_id = "mindlab-research/Macaron-V1-Preview-744B-Merged"

tokenizer = AutoTokenizer.from_pretrained(
    repo_id,
    trust_remote_code=True,
)

model = AutoModelForCausalLM.from_pretrained(
    repo_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)
model.eval()

Evaluation

For benchmark context and full Macaron-V1-Preview release notes, see the release blog and the routed 749B model card. This merged checkpoint is not equivalent to the full routed Mixture-of-LoRA serving stack, so behavior may differ from the 749B release.

License

Macaron-V1-Preview-744B-Merged is released under the MIT License. Users should also respect any requirements inherited from GLM-5.1 and from dependencies used in deployment.

Citation

@misc{mindlab2026macaronv1preview,
  author = {{Mind Lab}},
  title = {Macaron-V1-Preview: 749B MoL Agent Model post-trained from GLM5.1},
  year = {2026},
  howpublished = {Mind Lab: A Lab for Experiential Intelligence},
  note = {https://macaron.im/mindlab/research/macaron-v1-preview}
}

Contact

Organization: MindLab Research
Project: Macaron
Release blog: macaron.im/mindlab/research/macaron-v1-preview

Downloads last month: 51

Safetensors

Model size

754B params

Tensor type

BF16

F32

Model tree for mindlab-research/Macaron-V1-Preview-744B-Merged

Base model

zai-org/GLM-5.1

Finetuned

(11)

this model

Collection including mindlab-research/Macaron-V1-Preview-744B-Merged

Macaron-V1

Collection

5 items • Updated 20 days ago • 22