Instructions to use pipenetwork/Macaron-V1-Preview-749B-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use pipenetwork/Macaron-V1-Preview-749B-MLX-4bit with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("pipenetwork/Macaron-V1-Preview-749B-MLX-4bit")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use pipenetwork/Macaron-V1-Preview-749B-MLX-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use pipenetwork/Macaron-V1-Preview-749B-MLX-4bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default pipenetwork/Macaron-V1-Preview-749B-MLX-4bit

Run Hermes

hermes

MLX LM

How to use pipenetwork/Macaron-V1-Preview-749B-MLX-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "pipenetwork/Macaron-V1-Preview-749B-MLX-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Macaron-V1-Preview-749B-MLX-4bit

MLX (Apple Silicon) conversion of mindlab-research/Macaron-V1-Preview-749B — a 749B-parameter glm_moe_dsa (DeepSeek-V3.2-style sparse-attention MoE, 256 experts) — quantized to 4-bit. First MLX build of this model.

Quantizations

Part of the Macaron-V1-Preview-749B MLX collection.

Variant	Notes
4-bit (this repo)	4-bit · ~430GB · tight on 512GB
mixed	mixed · experts@3-bit, non-expert@6-bit · ~360GB · comfortable 512GB fit

The mixed build keeps the routed experts at 3-bit and the precision-sensitive non-expert layers (attention, shared experts, dense layers, embeddings, lm_head) at 6-bit, sized to run comfortably on a 512 GB Mac.

Use with mlx-lm

pip install mlx-lm
python -m mlx_lm generate --model pipenetwork/Macaron-V1-Preview-749B-MLX-4bit --prompt "Hello" -m 128

Validation

Smoke-tested locally (loads + generates coherent text).

License

MIT (inherited from base). Quantization config (excerpt): {"group_size": 64, "bits": 4, "mode": "affine"}.

Downloads last month: -

Safetensors

Model size

744B params

Tensor type

BF16

U32

F32

MLX

Hardware compatibility

4-bit

Model tree for pipenetwork/Macaron-V1-Preview-749B-MLX-4bit

Base model

zai-org/GLM-5.1

Finetuned

mindlab-research/Macaron-V1-Preview-749B

Quantized

(2)

this model

Collection including pipenetwork/Macaron-V1-Preview-749B-MLX-4bit

Macaron-V1-Preview-749B MLX

Collection

First MLX builds of Macaron-V1-Preview-749B (glm_moe_dsa, 749B MoE): a tight 4-bit and a 512GB-friendly mixed. • 2 items • Updated about 16 hours ago