Instructions to use jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k with Transformers:

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

pipe = pipeline("text-generation", model="jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k")
model = AutoModelForCausalLM.from_pretrained("jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k")
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 jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k

SGLang

How to use jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k 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 "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k" \
    --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": "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k",
		"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 "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k" \
        --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": "jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k with Docker Model Runner:
```
docker model run hf.co/jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k
```

GLM-5.2 NVFP4+AQLM hybrid (500k context variant)

The 524,288-token context variant of GLM-5.2-NVFP4-AQLM-hybrid for 4x 96 GB SM120 GPUs. The smaller KV cache and attention workspace fund 48% of experts at NVFP4 (4.5 bpw, covering ~68% of routed computations); the rest are 2-bpw AQLM. Weights: 334 GB. Trades context for more NVFP4 → higher quality than the 1M sibling at a smaller window.

🐳 Serve on SM120 (recommended)

Use the vLLM fork jarrelscy/vllm-glm52-sm120 — the hybrid NVFP4+AQLM MoE support, SM120 sparse-MLA (DSA) fixes, DCP, and both MTP / DSpark speculative decode are already committed (no runtime patching). The full per-config flag reference (with this variant's --max-model-len baked in) is in SERVING.md.

Decode throughput is set by the config, not the checkpoint — all variants decode identically per-config (validated on the 1M sibling: needle-in-haystack PASS @ 749K). The variants differ only in context ceiling; this checkpoint tops out near ~500k, so use MAXLEN=500000.

git clone -b glm52-sm120 https://github.com/jarrelscy/vllm-glm52-sm120
cd vllm-glm52-sm120
docker build -f Dockerfile.glm52-sm120 -t glm52-sm120 .

# ★ PREFERRED — ~500k window WITH lossless MTP spec, TP speed + graphs.
#   The tp4-1m-mtp entrypoint applies the 2026-07-13 promoted stack
#   automatically (IndexShare + bit-exact gemv kernels + ag_rs/NCCL-P2P
#   + chunk 4096 / util 0.97); MAXLEN caps the window at ~500k here.
docker run --gpus all --ipc=host -p 8001:8001 \
  -v /path/to/weights:/models/1m:ro \
  -e PARALLEL=tp4-1m-mtp -e MAXLEN=500000 glm52-sm120

# Fastest short-context decode, TP4 + DSpark:
docker run --gpus all --ipc=host -p 8001:8001 \
  -v /path/to/weights:/models/1m:ro -e PARALLEL=tp4-dspark glm52-sm120

Performance (measured, 4x RTX PRO 6000 SM120, 2026-07-13 promoted stack)

Single-stream decode (tp4-1m-mtp): short context ~77 tok/s counting / 71 code / 54 prose; at 123K context ~49 / 45 / 36 tok/s. Lossless (64K temp-0 golden-gated; reasoning canary PASS). This variant is slightly faster than the 1M sibling (73/64/52 short, 44/41/32 @123K) because its 48%-NVFP4 mix puts more experts on the fast path.
Fresh prefill: ~0.95K tok/s @123K (chunk 4096 + ag_rs/NCCL-P2P transport); prefix caching amortizes it across requests.

tp4-1m-mtp gives ~1M+spec on the 1M sibling; here the same config gives ~500k+spec (DCP — a stock vLLM feature for MLA models — shards the MLA latent KV by sequence; MTP's MLA-shaped draft KV shares that sharded latent). KV pool ~543K tokens at MAXLEN=500000. Override draft depth with -e NUM_SPEC=7 (structured/code) or -e NUM_SPEC=2 (general). Peel back any stack default with e.g. -e VLLM_MTP_INDEX_SHARE=0. The heavier NVFP4 weight per GPU leaves less KV room than the 1M variant; if a config OOMs, drop --max-model-len or --gpu-memory-utilization.

code/ in the main repo has the full production pipeline (routing stats, assignment solver, checkpoint builders) and SETUP.md. Quantized from lukealonso/GLM-5.2-NVFP4.

Downloads last month: 574

Safetensors

Model size

268B params

Tensor type

BF16

F32

F16

F8_E4M3

I16

Model tree for jarrelscy/GLM-5.2-NVFP4-AQLM-hybrid-500k

Base model

zai-org/GLM-5.2

Quantized

(110)

this model