---

🚀 v0.1.6: Real-time Metrics & Blackwell-Optimized Docker (Recommended)

This model is fully compatible with the DGX-Spark-llama.cpp-Bench. Experience the state-of-the-art inference engine optimized for NVIDIA Blackwell (DGX Spark) hardware.

🌟 Key Features (v0.1.6)

• Real-time Performance Metrics: Now visualizes Input TPS and Output TPS during streaming.
• Improved Reasoning UI: Seamlessly renders and stabilizes the model's Chain-of-Thought (CoT).
• Blackwell Optimization: Native support for ARM64/SM121 and CUDA 13.0 FP4.

🐳 Quick Start

# Pull the latest optimized image
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.6

For more details, visit our GitHub Repository.

---

🚀 v0.1.6: 실시간 지표 및 Blackwell 최적화 도커 (권장)

이 모델은 DGX-Spark-llama.cpp-Bench 시스템에 최적화되어 있습니다. NVIDIA Blackwell (DGX Spark) 하드웨어의 성능을 최대로 활용하세요.

🌟 주요 특징 (v0.1.6)

• 실시간 성능 지표 시각화: 스트리밍 중 Input TPS 및 Output TPS를 실시간으로 표시합니다.
• 지능형 추론 UI 고도화: 모델의 생각하는 과정(CoT)을 더 안정적으로 렌더링합니다.
• Blackwell 최적화: ARM64/SM121 아키텍처 및 CUDA 13.0 FP4 가속 지원.

🐳 실행 방법

# 최신 최적화 이미지 내려받기
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.6

상세한 사용법은 GitHub 리포지토리를 참조하세요.

---

---

🚀 v0.1.5: Real-time Metrics & Blackwell-Optimized Docker (Recommended)

This model is fully compatible with the DGX-Spark-llama.cpp-Bench. Experience the state-of-the-art inference engine optimized for NVIDIA Blackwell (DGX Spark) hardware.

🌟 Key Features (v0.1.5)

• Real-time Performance Metrics: Now visualizes Input TPS and Output TPS during streaming.
• Improved Reasoning UI: Seamlessly renders and stabilizes the model's Chain-of-Thought (CoT).
• Blackwell Optimization: Native support for ARM64/SM121 and CUDA 13.0 FP4.

🐳 Quick Start

# Pull the latest optimized image
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.5

For more details, visit our GitHub Repository.

---

🚀 v0.1.5: 실시간 지표 및 Blackwell 최적화 도커 (권장)

이 모델은 DGX-Spark-llama.cpp-Bench 시스템에 최적화되어 있습니다. NVIDIA Blackwell (DGX Spark) 하드웨어의 성능을 최대로 활용하세요.

🌟 주요 특징 (v0.1.5)

• 실시간 성능 지표 시각화: 스트리밍 중 Input TPS 및 Output TPS를 실시간으로 표시합니다.
• 지능형 추론 UI 고도화: 모델의 생각하는 과정(CoT)을 더 안정적으로 렌더링합니다.
• Blackwell 최적화: ARM64/SM121 아키텍처 및 CUDA 13.0 FP4 가속 지원.

🐳 실행 방법

# 최신 최적화 이미지 내려받기
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.5

상세한 사용법은 GitHub 리포지토리를 참조하세요.

---

---

🚀 v0.1.4: Quick Start with Blackwell-Optimized Docker (Recommended)

This model is fully compatible with the DGX-Spark-llama.cpp-Bench. Experience the best performance on NVIDIA Blackwell (DGX Spark) hardware with our optimized inference engine.

🌟 Key Features (v0.1.4)

• Blackwell Optimized: Native support for ARM64/SM121 and CUDA 13.0 FP4.
• Intelligent Reasoning UI: Automatic extraction and visualization of reasoning processes (CoT).
• One-Click Deployment: Standardized environment via GHCR Docker image.

🐳 How to Run

# Pull the latest optimized image
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.4

# Follow the instructions in our repo to serve this model
# GitHub: https://github.com/sowilow/DGX-Spark-llama.cpp-Bench

---

🚀 v0.1.4: Blackwell 최적화 도커 퀵스타트 (권장)

이 모델은 DGX-Spark-llama.cpp-Bench 시스템에 최적화되어 있습니다. NVIDIA Blackwell (DGX Spark) 하드웨어의 성능을 최대로 활용하는 최적화된 추론 엔진을 경험해 보세요.

🌟 주요 특징 (v0.1.4)

• Blackwell 최적화: ARM64/SM121 아키텍처 및 CUDA 13.0 FP4 하드웨어 가속 지원.
• 지능형 추론 UI: 모델의 생각하는 과정(CoT)을 자동으로 감지하고 시각화합니다.
• 간편한 배포: GHCR 도커 이미지를 통해 환경 설정 없이 즉시 실행 가능합니다.

🐳 실행 방법

# 최신 최적화 이미지 내려받기
docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:v0.1.4

상세한 사용법은 GitHub 리포지토리를 참조하세요.

---

---

🚀 Quick Start with Docker (Recommended)

You can easily run this model using the DGX-Spark-llama.cpp-Bench inference engine. It's pre-configured for high-performance inference on NVIDIA hardware (especially Blackwell/DGX Spark).

1. Pull the Docker Image

docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:latest

2. Run the Inference Server

For detailed configuration and usage, visit the GitHub Repository.

---

gpt-oss-20b-DGX-Spark-GGUF

This repository provides GGUF quantized versions of OpenAI's gpt-oss-20b, optimized specifically for NVIDIA Blackwell (DGX Spark) architectures.

These models were converted and quantized using llama.cpp with support for the gpt_oss architecture.

Model Highlights

• Optimized for Blackwell: Specifically tuned for high-performance inference on NVIDIA DGX Spark (SM120/SM121).
• Flexible Quantization:
  • Q4_MXFP4: 4-bit Medium quantization (recommended for efficiency).
  • Q8_0: 8-bit quantization (recommended for maximum precision).
• MoE Architecture: 21B total parameters with 3.6B active parameters, leveraging Mixture-of-Experts for high efficiency.
• Long Context: Supports up to 131k context length.

Quantization Details

File	Quant Method	Bitrate	Size	Description
gpt-oss-20b-q4_mxfp4.gguf	Q4_MXFP4	4.5 bpw	~12 GB	Balanced performance and quality.
gpt-oss-20b-q8_0.gguf	Q8_0	8.5 bpw	~22 GB	Standard 8-bit quantization.

Quick Start (llama.cpp)

To run these models on a DGX Spark system:

1. Pull the optimized Docker image:

   docker pull ghcr.io/sowilow/dgx-spark-llama.cpp-bench:latest
   

2. Run with llama-server:

   docker run --gpus all -v $(pwd)/models:/model \
       ghcr.io/sowilow/dgx-spark-llama.cpp-bench:latest \
       llama-server -m /model/gpt-oss-20b-q4_mxfp4.gguf -ngl 99 -c 8192
   

Original Model Information

This is a quantized version of openai/gpt-oss-20b. Please refer to the original model card for details on training, safety, and benchmarks.

Citation

@misc{openai2025gptoss120bgptoss20bmodel,
      title={gpt-oss-120b & gpt-oss-20b Model Card}, 
      author={OpenAI},
      year={2025},
      eprint={2508.10925},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2508.10925}, 
}