{
  "status": "healthy",
  "redis": {"connected": true, "latency_ms": 0.5, "used_mb": 1.8, "limit_mb": 256, "usage_pct": 0.7}
}

{
  "status": "healthy",                 // healthy / degraded / unhealthy / idle
  "gpu_relay": true,
  "inference_servers": {"tei": {"healthy": true, "status_code": 200, ...}, ...},
  "tasks": {
    "embedding": {"enabled": true, "worker_alive": true, "client_connected": true,
                  "inference_server": "tei", "inference_server_healthy": true},
    ...
  }
}

// 모든 활성 task ready → 200
{"ready": true, "gpu_relay": true, "reason": "ready", "draining": false,
 "tasks": {"translation": {"ready": true, "reason": "ready"}, ...}}

// 일부 task 워밍업 중 → 503 (LB가 트래픽 보내면 안 됨)
{"ready": false, "gpu_relay": true, "reason": "warming up", "draining": false,
 "tasks": {"stt": {"ready": false, "reason": "warming_up"}, ...}}

// drain 모드 (SIGTERM 후) → 503 (LB가 트래픽 빼야 함)
{"ready": false, "gpu_relay": true, "reason": "draining", "draining": true, ...}

{
  "object": "list",
  "data": [
    {"id": "BAAI/bge-m3", "object": "model", "created": 0, "owned_by": "redgx", "task": "embedding"},
    {"id": "LGAI-EXAONE/EXAONE-4.0-1.2B", "object": "model", "owned_by": "redgx", "task": "generation"}
  ],
  "redgx": {
    "gpu_relay": true,
    "draining": false,
    "tasks": {
      "embedding": {
        "enabled": true,
        "model": "BAAI/bge-m3",
        "inference_server": {"id": "tei", "base_url": "http://gpu-tei-embedding:8080"},
        "worker": {"running": true, "paused": false, "ready": true, "reason": "ready"},
        "queue_depth": 0,
        "inflight": 0,
        "circuit_breaker": {"state": "closed", "failure_count": 0, "trips_total": 0, ...}
      },
      "generation": {"enabled": false}, ...
    }
  }
}

# HELP redgx_requests_total Total requests by HTTP status
# TYPE redgx_requests_total counter
redgx_requests_total{status="200"} 1763
redgx_request_duration_ms_bucket{le="5"} 695
redgx_redis_memory_bytes 1878296
...

# Counter
redgx_gpu_submitted_total{task="embedding"} 1234
redgx_gpu_completed_total{task="embedding"} 1230
redgx_gpu_inference_errors_total{task="embedding",kind="timeout"} 2

# Gauge (실시간)
redgx_gpu_queue_depth{task="embedding"} 5
redgx_gpu_inflight{task="embedding"} 8

# Histogram (배치 크기)
redgx_gpu_batch_size_bucket{task="embedding",le="8"} 100
redgx_gpu_batch_size_sum{task="embedding"} 450
redgx_gpu_batch_size_count{task="embedding"} 100

# Histogram (추론 지연 — task × server 라벨)
redgx_gpu_inference_latency_seconds_bucket{task="embedding",server="tei",le="0.5"} 50
redgx_gpu_inference_latency_seconds_sum{task="embedding",server="tei"} 12.3
redgx_gpu_inference_latency_seconds_count{task="embedding",server="tei"} 100

// 요청
{
  "model": "LGAI-EXAONE/EXAONE-4.0-1.2B",
  "messages": [
    {"role": "system", "content": "당신은 한국어 비서입니다."},
    {"role": "user", "content": "파이썬과 자바의 차이점을 한 문장으로"}
  ],
  "max_tokens": 256,
  "temperature": 0.7
}

// 응답 (200)
{
  "id": "chatcmpl-xxx",
  "object": "chat.completion",
  "created": 1714200000,
  "model": "LGAI-EXAONE/EXAONE-4.0-1.2B",
  "choices": [{"index": 0, "message": {"role": "assistant", "content": "..."}, "finish_reason": "stop"}],
  "usage": {"prompt_tokens": 30, "completion_tokens": 50, "total_tokens": 80}
}

메시지 매핑 규칙

• role: system → RedGX inbox system_prompt (다중이면 \n\n 결합)
• role: user / assistant → chat-style 직렬화 (User: ...\nAssistant: ...)로 prompt에 합침
• 마지막에 Assistant:를 붙여 모델 응답 유도

지원/비지원 필드

// 요청
{
  "model": "BAAI/bge-m3",
  "input": ["문장 1", "문장 2"],     // str 또는 list[str]
  "encoding_format": "float"        // "float"(기본) 또는 "base64"
}

// 응답 (200)
{
  "object": "list",
  "data": [
    {"object": "embedding", "index": 0, "embedding": [0.123, -0.456, ...]},
    {"object": "embedding", "index": 1, "embedding": [0.789, ...]}
  ],
  "model": "BAAI/bge-m3",
  "usage": {"prompt_tokens": 0, "total_tokens": 0}    // RedGX는 토큰 카운트 미제공
}

// 요청
{
  "model": "LGAI-EXAONE/EXAONE-4.0-1.2B",
  "messages": [{"role": "user", "content": "파이썬 hello world"}],
  "stream": true
}

// 응답 (text/event-stream)
data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant","content":""}}]}

data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"print"}}]}

data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"(\"hello\")"}}]}

data: {"id":"...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]