Instructions to use alibaybay/MiniMax-H3-Prompt-Rewriter-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 alibaybay/MiniMax-H3-Prompt-Rewriter-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 alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

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 alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

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 alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

Use Docker

docker model run hf.co/alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "alibaybay/MiniMax-H3-Prompt-Rewriter-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": "alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

Ollama
How to use alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF with Ollama:
```
ollama run hf.co/alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF with Docker Model Runner:
```
docker model run hf.co/alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M
```

Lemonade

How to use alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.MiniMax-H3-Prompt-Rewriter-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

MiniMax-H3 Prompt Rewriter (Qwen2.5-Omni-7B + LoRA) - GGUF

This repository contains the quantized GGUF weights for MiniMax-H3 Prompt Rewriter, created by merging the LoRA adapter lightx2v/MiniMax-H3-Prompt-Rewriter-LoRA-Omni with Qwen/Qwen2.5-Omni-7B (Thinker backbone).

Quantized using llama.cpp on Modal.com GPU infrastructure.

Available Quantizations

File	Size	Description	Recommended For
MiniMax-H3-Prompt-Rewriter-Q4_K_M.gguf	~4.5 GB	4-bit medium quantization	CPU & low-VRAM GPUs (Ollama)
MiniMax-H3-Prompt-Rewriter-Q8_0.gguf	~7.5 GB	8-bit high-fidelity quantization	Near-FP16 accuracy

Quick Start with Ollama

You can run this model directly via Ollama:

ash ollama run hf.co/alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF:Q4_K_M

Or using the included Modelfile: ash ollama create minimax-h3-rewriter -f ./Modelfile ollama run minimax-h3-rewriter

Prompt Rewriting Schema

The model restructures short user requests into the official MiniMax-H3 video & audio generation schema: - integrated_multimodal_description: Shot breakdown, camera motions, and visual styling. - overall_soundscape: Ambient sound, physical foley, non-verbal sound.

on_diegetic_music: Instrumentation, tempo, and background score.

Downloads last month: 277

GGUF

Model size

8B params

Architecture

qwen2

Hardware compatibility

4-bit

8-bit

Model tree for alibaybay/MiniMax-H3-Prompt-Rewriter-GGUF

Base model

Qwen/Qwen2.5-Omni-7B

Quantized

(24)

this model