Instructions to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
# Run inference directly in the terminal:
llama cli -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
# Run inference directly in the terminal:
llama cli -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
# Run inference directly in the terminal:
./llama-cli -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
# Run inference directly in the terminal:
./build/bin/llama-cli -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

Use Docker

docker model run hf.co/PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

LM Studio
Jan
Ollama
How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF with Ollama:
```
ollama run hf.co/PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
```
Unsloth Desktop

How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

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": "PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF with Docker Model Runner:
```
docker model run hf.co/PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF
```

Lemonade

How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

Run and chat with the model

lemonade run user.GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

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 PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF

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 "PaoAI/GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED-GGUF" \
  --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"

What lamma fork to use

by stevilg - opened 1 day ago

Discussion

stevilg

1 day ago

•

edited 1 day ago

That has both glm support and rcomfp?

The ROCmFPX fp6 Vulkan build (v140, commit 647fd52) loads the file but reports: unknown model architecture: 'glm5next' │
We also have the upstream Unsloth llama.cpp fork with the glm5next/upstream branch (commit d94f44e79) which implements LLM_ARCH_GLM5NEXT, but it fails with: tensor 'blk.0.ffn_down.weight' has invalid ggml type 101. should be in [0, 43) — this is because the model uses GGML_TYPE_Q4_0_ROCMFP4_FAST (type 101), which is a ROCmFPX custom quantization type not present in upstream.

PaoAI

Owner 1 day ago

Use kingjones30/ROCmFPX, branch main, commit 3345156 or later — that's the exact tree we build and serve this model with right now.

Your two errors are explained by the two halves of the puzzle:

Build	What it has	What it lacks	Result
ROCmFPX v140 (647fd52)	`Q4_0_ROCMFP4_FAST` (type 101)	glm5next arch	`unknown model architecture: 'glm5next'`
Unsloth glm5next branch (d94f44e79)	glm5next arch	our custom FP4 type	`invalid ggml type 101`

kingjones30's main after the glm5next merge (3345156) has both.

Build with Vulkan:

cmake -B build -DGGML_VULKAN=ON && cmake --build build --config Release -j

Then (this is our exact serving command, running now):

llama-server -m GLM-5.3-Flash-PaoAI-ROCmFP4-STRIX-BALANCED.gguf \
  --host 0.0.0.0 --port 8080 -ngl 999 -c 65536 --parallel 1 \
  --spec-type draft-mtp --spec-draft-n-max 4 \
  -fa on --cache-type-k q8_0 --cache-type-v q8_0 --lazy-mode off

On a 128GB Strix Halo this runs full-GPU with no offload flags. We've also added a Requirements section to the model card documenting this.

(If your GPU is gfx1151 / Ryzen AI Max 395 this should just work — that's everything we've tested.)

bongole

about 15 hours ago

Hi.
I can't find commit 3345156 in kingjones30/ROCmFPX — it's not on any branch or tag. Is it maybe not pushed yet?

PaoAI

Owner about 14 hours ago

You're right — sorry about that. 3345156 was our local merge (Unsloth glm5next/upstream arch into ROCmFPX) and it had never been pushed anywhere public. Our mistake, thanks for catching it.

It's public now:

Fork: https://github.com/guevae2/ROCmFPX
Branch: glm5next
Commit: 3345156 — head of that branch

Build with Vulkan (cmake -B build -DGGML_VULKAN=ON) and the serving command from the model card works as documented on gfx1151 / Ryzen AI Max 395. We've also updated the model cards to point at the fork instead of the unreachable reference.

(If kingjones30 upstreams glm5next later, main will work too — we'll note it here.)

PaoAI

Owner about 14 hours ago

Also a thanks to @stevilg — you asked the question that surfaced this in the first place. Both the build path and the card references are fixed now; if either of you hits anything else running these models, this thread is the right place.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment