Instructions to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
# Run inference directly in the terminal:
llama cli -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
# Run inference directly in the terminal:
llama cli -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
# Run inference directly in the terminal:
./llama-cli -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
# Run inference directly in the terminal:
./build/bin/llama-cli -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Use Docker

docker model run hf.co/xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Llama new
LM Studio
Jan

vLLM

How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "xCloudinfo/gpt-oss-120b-TAIDE-zhTW-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": "xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Ollama
How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with Ollama:
```
ollama run hf.co/xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
```
Unsloth Desktop

How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

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": "xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with Docker Model Runner:
```
docker model run hf.co/xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4
```

Lemonade

How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Run and chat with the model

lemonade run user.gpt-oss-120b-TAIDE-zhTW-GGUF-MXFP4

List all available models

lemonade list

Hermes Agent

How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

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 xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4

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 "xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF:MXFP4" \
  --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"

gpt-oss-120b-TAIDE-zhTW-GGUF

云碩科技 · xCloudinfo　·　系列：繁中在地化 · TAIDE zh-TW

繁體中文（台灣）reasoning大模型的GGUF。以 openai/gpt-oss-120b（117B總參 / 5.1B活躍 / 128-expert MoE / MXFP4 / harmony推理格式）為基底，用 TAIDE 蒸餾的台灣繁中self-instruct指令資料做 LoRA微調（LoRA作用於attention，MoE專家維持原生MXFP4），保留gpt-oss原生reasoning能力。

做法

資料：以TAIDE蒸餾的台灣繁中self-instruct指令資料訓練。
訓練：於 云碩AI算力資源池（云碩算力中心） 上以分散式LoRA微調，合併回完整模型後轉為GGUF。資料全程留在自有算力環境、流程可重現。
格式：MXFP4 GGUF（≈65GB），llama.cpp / Ollama直接載入。
效果：相較原始gpt-oss，本模型直接以道地台灣繁中作答，用字採台灣慣用（臺灣、臺語、客家話），語氣更在地。

用法

gpt-oss是reasoning模型，使用 harmony chat模板（GGUF內已內建），請套用內建模板並給足生成長度。

llama.cpp

下載：

hf download xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF TAIDE-gpt-oss-120b-MXFP4.gguf --local-dir .

OpenAI相容服務（推薦，--jinja 套用內建harmony模板）：

llama-server -m TAIDE-gpt-oss-120b-MXFP4.gguf --jinja -ngl 999 -c 4096 --host 0.0.0.0 --port 8080

一次性對話：

llama-cli -m TAIDE-gpt-oss-120b-MXFP4.gguf --jinja -ngl 999 -c 4096 \
  -p "用一段話介紹台灣，並說說台灣最有名的小吃。"

Ollama

cat > Modelfile <<'EOF'
FROM ./TAIDE-gpt-oss-120b-MXFP4.gguf
PARAMETER num_ctx 4096
EOF
ollama create gpt-oss-120b-taide-zhtw -f Modelfile
ollama run gpt-oss-120b-taide-zhtw "請用繁體中文解釋什麼是遞迴，並給一個生活化的比喻。"

注意：本模型會先在reasoning階段思考再輸出最終答案，請給足回覆長度，否則可能只看到思考、看不到答案。

完整safetensors（transformers / vLLM）版本另見對應repo。

授權與來源聲明

基底：openai/gpt-oss-120b，Apache-2.0。
TAIDE衍生：訓練資料以 TAIDE 模型蒸餾；使用/重製/改作/散布須遵守 TAIDE模型授權條款並保留本聲明。
不得用於軍事或非法用途；不得聲稱代表TAIDE或國家實驗研究院；須遵守中華民國法律與適用之EU AI Act。

由云碩科技xCloudinfo於自有AI算力資源池製作；資料留在本地、流程可重現。

關於云碩科技

云碩科技（xCloudinfo Corp. Limited）專注於地端私有化人工智慧解決方案。xVITA是云碩自研模型的品牌名稱。

其他模型請見 https://hf.co/xCloudinfo。

gpt-oss-120b-TAIDE-zhTW-GGUF — English

xCloudinfo (云碩科技)　·　Series: Traditional-Chinese localization · TAIDE zh-TW

The GGUF of a Traditional Chinese (Taiwan) reasoning large model. Based on openai/gpt-oss-120b (117B total parameters / 5.1B active / 128-expert MoE / MXFP4 / harmony reasoning format), fine-tuned via LoRA on Taiwanese-Chinese self-instruct instruction data distilled from TAIDE (LoRA acts on attention; the MoE experts stay native MXFP4), preserving gpt-oss's native reasoning ability.

Approach

Data: trained on Taiwanese-Chinese self-instruct instruction data distilled from TAIDE.
Training: distributed LoRA fine-tuning on the xCloudinfo AI compute resource pool (xCloudinfo compute center), merged back into the full model and then converted to GGUF. The data stays within the in-house compute environment throughout, and the process is reproducible.
Format: MXFP4 GGUF (≈65GB), loaded directly by llama.cpp / Ollama.
Effect: compared with the original gpt-oss, this model answers directly in idiomatic Taiwanese Chinese, with Taiwanese usage (臺灣, 臺語, 客家話) and a more local tone.

Usage

gpt-oss is a reasoning model and uses the harmony chat template (already built into the GGUF); apply the built-in template and give it enough generation length.

llama.cpp

Download:

hf download xCloudinfo/gpt-oss-120b-TAIDE-zhTW-GGUF TAIDE-gpt-oss-120b-MXFP4.gguf --local-dir .

OpenAI-compatible service (recommended, --jinja applies the built-in harmony template):

llama-server -m TAIDE-gpt-oss-120b-MXFP4.gguf --jinja -ngl 999 -c 4096 --host 0.0.0.0 --port 8080

One-shot conversation:

llama-cli -m TAIDE-gpt-oss-120b-MXFP4.gguf --jinja -ngl 999 -c 4096 \
  -p "用一段話介紹台灣，並說說台灣最有名的小吃。"

Ollama

cat > Modelfile <<'EOF'
FROM ./TAIDE-gpt-oss-120b-MXFP4.gguf
PARAMETER num_ctx 4096
EOF
ollama create gpt-oss-120b-taide-zhtw -f Modelfile
ollama run gpt-oss-120b-taide-zhtw "請用繁體中文解釋什麼是遞迴，並給一個生活化的比喻。"

Note: this model thinks first in a reasoning stage and then outputs the final answer, so give it enough reply length, otherwise you may see only the thinking and not the answer.

For the full safetensors (transformers / vLLM) version, see the corresponding repo.

License & provenance

Base: openai/gpt-oss-120b, Apache-2.0.
TAIDE derivative: the training data is distilled from the TAIDE model; use/reproduction/adaptation/redistribution must comply with the TAIDE model license terms and retain this notice.
Must not be used for military or illegal purposes; must not claim to represent TAIDE or the National Applied Research Laboratories; must comply with the laws of the Republic of China (Taiwan) and the applicable EU AI Act.

Made by xCloudinfo (云碩科技) on its own AI compute resource pool; data stays local, process reproducible.