Instructions to use pipenetwork/GLM-5.2-REAP37-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use pipenetwork/GLM-5.2-REAP37-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/GLM-5.2-REAP37-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/GLM-5.2-REAP37-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/GLM-5.2-REAP37-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/GLM-5.2-REAP37-MLX-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use pipenetwork/GLM-5.2-REAP37-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/GLM-5.2-REAP37-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/GLM-5.2-REAP37-MLX-4bit

Run Hermes

hermes

OpenClaw new

How to use pipenetwork/GLM-5.2-REAP37-MLX-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "pipenetwork/GLM-5.2-REAP37-MLX-4bit"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "pipenetwork/GLM-5.2-REAP37-MLX-4bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

MLX LM

How to use pipenetwork/GLM-5.2-REAP37-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/GLM-5.2-REAP37-MLX-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "pipenetwork/GLM-5.2-REAP37-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/GLM-5.2-REAP37-MLX-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

GLM-5.2-REAP37-MLX-4bit

REAP expert-pruned + 4-bit MLX conversion of zai-org/GLM-5.2. Keeps the 160 most-salient experts per layer (of 256) → ~480B params, smaller/faster than the full model.

What is this

Pruned with REAP (Router-weighted Expert Activation Pruning, Cerebras / ICLR 2026): per MoE layer, experts are scored by mean(router_gate_weight × ‖expert_output‖) over a calibration set; the lowest-saliency experts are dropped and the router is sliced to the survivors. No retraining. n_routed_experts reduced 256→160.

Quality (held-out perplexity, Frankenstein — not in calibration)

Variant	Experts	~Params	Held-out PPL	vs full
full GLM-5.2 (4-bit)	256	~750B	1.447	—
REAP25	192	~572B	1.481	+2.3%
REAP37 (this repo)	160	~480B	1.553	+7.3%
REAP50	128	~394B	1.990	+37.5%

This variant: PPL 1.553 (+7.3% vs full) — modest quality cost. (Absolute PPL is low because the eval text is highly predictable; treat the numbers as relative degradation.)

Methodology

Calibrated on the 4-bit GLM-5.2 (192 seqs × 1024 tok, prose + code); pruned during MLX conversion (no intermediate bf16). Requires the glm_moe_dsa / deepseek_v32 MLX path with per-layer indexer handling.

Use with mlx-lm

pip install mlx-lm
python -m mlx_lm generate --model pipenetwork/GLM-5.2-REAP37-MLX-4bit --prompt "Hello" -m 256

License

MIT (inherited from GLM-5.2). Quantization: {"group_size": 64, "bits": 4, "mode": "affine"}.

Downloads last month: 3,433

Safetensors

Model size

472B params

Tensor type

BF16

U32

F32

MLX

Hardware compatibility

4-bit

Model tree for pipenetwork/GLM-5.2-REAP37-MLX-4bit

Base model

zai-org/GLM-5.2

Quantized

(110)

this model

Collection including pipenetwork/GLM-5.2-REAP37-MLX-4bit

GLM-5.2 REAP (expert-pruned)

Collection

REAP expert-pruned GLM-5.2 (smaller/faster). REAP-25 near-lossless (+2.3% PPL); REAP-37 +7.3%; REAP-50 smallest but +37.5%. • 3 items • Updated about 1 month ago