examples/runtime_quickstart.md

# Runtime quickstart

## 1. Start Qwen3-8B

```bash
python -m vllm.entrypoints.openai.api_server \
  --model Qwen/Qwen3-8B \
  --served-model-name qwen3-8b-base \
  --dtype bfloat16 \
  --max-model-len 2048 \
  --gpu-memory-utilization 0.50 \
  --max-num-seqs 8
```

## 2. Clone this repo

```bash
git clone https://huggingface.co/koreallmdev/qwen3-8b-korean-coding-router
cd qwen3-8b-korean-coding-router
```

## 3. Run router CLI

```bash
export OPENAI_BASE_URL=http://localhost:8000/v1
export OPENAI_MODEL=qwen3-8b-base

bash scripts/run_qwen3_8b_korean_coding_router_chat.sh
```

## 4. Test

```text
/cat linux 8010 포트가 이미 사용 중일 때 ss -lntp와 ps를 포함해서 확인 순서 알려줘
/cat longform 주제는 코딩이고 4천자 이상 텍스트를 써줘
```