Instructions to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
# Run inference directly in the terminal:
llama cli -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
# Run inference directly in the terminal:
llama cli -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
# Run inference directly in the terminal:
./llama-cli -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Use Docker

docker model run hf.co/pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

LM Studio
Jan

vLLM

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Ollama
How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with Ollama:
```
ollama run hf.co/pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
```
Unsloth Desktop

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with Docker Model Runner:
```
docker model run hf.co/pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16
```

Lemonade

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Run and chat with the model

lemonade run user.GLM-5.3-Flash-GSQ-RCO-GGUF-BF16

List all available models

lemonade list

Hermes Agent

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

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 pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16

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 "pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF:BF16" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

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

TL;DR: GLM's 320B-parameter multimodal model in 118.6 to 138.2 GB including vision, compared with 328.3 GB for the FP8 checkpoint.

GLM-5.3-Flash · GSQ-RCO GGUFs

Non-uniform GGUF quantizations produced with GSQ and RCO, with a vision projector for multimodal use.

Independent community reproduction. These files were produced by a third party using the published GSQ and RCO methods. They are not an IST-DASLab release and carry no endorsement from the authors of either paper.

Overview

This repository contains two GGUF quantizations of zai-org/GLM-5.3-Flash and its BF16 vision projector (mmproj). RCO assigns each weight tensor a quantization type according to its sensitivity, within an exact total size budget.

Method	Description
GSQ (Gumbel-Softmax Quantization, paper, code)	Post-training scalar quantization that jointly learns per-coordinate grid assignments and per-group scales through a Gumbel-Softmax relaxation.
RCO (Riemannian Constrained Optimization, paper, code)	Assigns one of K quantization types to each of N tensors under an exact total size budget, reformulated as a smooth Riemannian manifold in logit space.

Both methods were developed at the Deep Algorithms and Systems Lab (DASLab), Institute of Science and Technology Austria.

Read this before using these files

Evaluation uses llama.cpp PR 27773 with native-f32-mmf.patch and NVIDIA_TF32_OVERRIDE=0 GGML_CUDA_MMF_F32_DISABLE=1. The runtime instructions include the build and launch commands.

Available files

File	bpw	Size	Notes
`GLM-5.3-Flash-GSQ-RCO-3.5bit.gguf`	3.499816	137.07 GB	Better perplexity of the two
`GLM-5.3-Flash-GSQ-RCO-3.0bit.gguf`	2.999595	117.48 GB	Smaller
`GLM-5.3-Flash-mmproj-BF16.gguf`	16.52	1.16 GB	Vision encoder and projector

MMLU-Pro

No reasoning; 2,048-token context limit. MMLU-Pro uses 2,000 questions sampled with a fixed seed and stratified across all 14 categories. Questions are scored zero-shot using the log probabilities of single-token answers A through J, without a chat template. All three builds use the same patched runtime and matching layer placement. Raw scores: Q8_0, 3.5-bit, 3.0-bit.

The eight-bit Q8_0 reference is converted from the pinned FP8 source, excluding the unused MTP layer. The reference record contains the conversion details and hashes.

Build	Accuracy	SE	vs Q8_0, paired	Discordant (Q8_0 right / build right)	Exact p
Q8_0 reference	61.95%	1.09
GSQ-RCO 3.5-bit	60.55%	1.09	-1.40 pp	86 / 58	0.0241
GSQ-RCO 3.0-bit	60.00%	1.10	-1.95 pp	128 / 89	0.00973

Correct counts are 1,239, 1,211 and 1,200 out of 2,000. Chance is 11.2%. The 3.5-bit model leads 3.0-bit by 0.55 pp (118 / 107 discordant, exact p = 0.505). Q8_0 and 3.5-bit agree on 1,739 predictions. P-values are two-sided and unadjusted.

Per category:

Category	n	Q8_0	3.5-bit	3.0-bit
biology	119	93.3%	93.3%	88.2%
business	131	48.1%	46.6%	48.1%
chemistry	188	45.2%	44.7%	41.5%
computer science	68	76.5%	76.5%	75.0%
economics	140	80.7%	80.0%	79.3%
engineering	161	49.7%	48.4%	52.8%
health	136	71.3%	66.9%	72.1%
history	63	73.0%	74.6%	71.4%
law	183	58.5%	57.9%	56.3%
math	225	49.8%	44.9%	44.0%
other	154	69.5%	68.8%	65.6%
philosophy	83	66.3%	69.9%	71.1%
physics	216	45.4%	42.6%	42.1%
psychology	133	85.0%	84.2%	83.5%

Results

Perplexity is measured on the same held-out text for all three builds: eight 1,024-token contexts and 4,088 scored tokens. Generation scores cover 16 IFEval and 8 GSM8K items. Completed-correct counts exclude responses that hit the token cap.

Variant	bpw	GB	Native PPL↓	vs Q8_0	Approx. KL↓	IFEval strict	IFEval completed correct	GSM8K completed correct
Q8_0 reference	n/a	n/a	3.3909	n/a	n/a	4/16	2/16	5/8
GSQ-RCO 3.5-bit	3.499816	137.07	3.5431	+4.49%	0.071346	7/16	4/16	5/8
GSQ-RCO 3.0-bit	2.999595	117.48	3.6985	+9.07%	0.142370	10/16	4/16	5/8

Q8_0 emitted reasoning on five IFEval requests despite the thinking-off adapter. KL is approximate because the native reference cache stores quantized log probabilities. Raw responses: Q8_0, 3.5-bit, 3.0-bit.

Runtime

The base is upstream 8134115f.
llama.cpp PR 27773 adds support for the glm5-next architecture.
native-f32-mmf.patch from this repository: rejects F32 candidates in the MMF path so no TF32 tensor-core instructions are emitted. NVIDIA_TF32_OVERRIDE=0 alone does not do this. No behaviour changes unless GGML_CUDA_MMF_F32_DISABLE=1 is set.

git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp
git fetch origin de25343596a2c924ba62d35b126e811be0d38a49
git checkout de25343596a2c924ba62d35b126e811be0d38a49
git apply /path/to/native-f32-mmf.patch
cmake -S . -B build -DGGML_CUDA=ON -DCMAKE_BUILD_TYPE=Release
cmake --build build --target llama-server -j

NVIDIA_TF32_OVERRIDE=0 GGML_CUDA_MMF_F32_DISABLE=1 ./build/bin/llama-server \
  -m GLM-5.3-Flash-GSQ-RCO-3.5bit.gguf \
  --mmproj GLM-5.3-Flash-mmproj-BF16.gguf \
  -ngl all -c 16384 -np 4 --kv-unified --load-mode none --lazy-mode off -fa off --jinja

hf download pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF \
  GLM-5.3-Flash-GSQ-RCO-3.5bit.gguf GLM-5.3-Flash-mmproj-BF16.gguf --local-dir .

Quantization procedure

GSQ trained expert projection candidates across 45 layers and 288 experts for 16 steps, using 64 training and 4 validation sequences of 1024 tokens.
RCO selected the allocation by validation teacher KL. Neither target improved on step 0, so both kept their initial allocation.
The size budget included packed weights, metadata and alignment. A roundtrip check verified the payload.

Source revision: eb9eb208.

Target	Q4_K	Q3_K	Q2_K
3.5-bit	2	105	19
3.0-bit	9	9	108

Citation

If you use these files, please cite this release together with the base model and both methods.

This release

@misc{glm53flashgsqrco2026,
  title        = {GLM-5.3-Flash GSQ-RCO GGUF quantizations},
  author       = {Josephine Pfeiffer},
  year         = {2026},
  publisher    = {Hugging Face},
  doi          = {10.57967/hf/10400},
  howpublished = {\url{https://huggingface.co/pfeifferj/GLM-5.3-Flash-GSQ-RCO-GGUF}}
}

Base model

@misc{glm53flash,
  title  = {GLM-5.3-Flash},
  author = {{Zhipu AI}},
  year   = {2026},
  url    = {https://huggingface.co/zai-org/GLM-5.3-Flash}
}

Methods

@article{gsq2026,
  title  = {GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling},
  author = {Dadgarnia, Alireza and Tabesh, Soroush and Nikdan, Mahdi and Helcig, Michael and Kurtic, Eldar and Kleinegger, Maximilian and Alistarh, Dan},
  journal= {arXiv preprint arXiv:2604.18556},
  year   = {2026}
}
@article{rco2026,
  title  = {Model Compression with Exact Budget Constraints via Riemannian Manifolds},
  author = {Helcig, Michael and Alistarh, Dan},
  journal= {arXiv preprint arXiv:2605.00649},
  year   = {2026}
}

Acknowledgements

Huge kudos to the Deep Algorithms and Systems Lab (DASLab) at the Institute of Science and Technology Austria for developing GSQ and RCO and for releasing the papers and reference implementations publicly.

Thanks also to ella for giving me the idea and supporting me at every step of the process, and Chris van Hoof (Red Hat) for contributing the compute on which these quantizations were produced and evaluated.

License

These quantized weights inherit the license of the base model, zai-org/GLM-5.3-Flash (MIT); the original license text ships as LICENSE. The GSQ and RCO tooling is released by DASLab under its own repository licenses.