{
  "updated": "2026-10-09",
  "route_version": "2026-09-24",
  "route_page": "route-2026-09-24.html",
  "updated_note": "2026-10-09：语音识别换 SenseVoice，funasr / faster-whisper-medium / sensevoice 同一引擎，见 asr.html。2026-10-06 晚：演播按 clavue-tts-ml/fn/rest 分组；acted 网关 8 分片，:31801 已停；luxtts 2 GPU；ASMR custom+lane2。",
  "gateway_doc": "gateway-upstream-2026-10-06.html",
  "serving_2b": {
    "deploy": "vllm-vibecode-2b-v38-004",
    "tensor_parallel": 1,
    "replicas": 6,
    "max_model_len": 262144,
    "max_num_seqs_per_replica": 24,
    "full_context_concurrency_per_replica": 4.02,
    "kv_cache_tokens_per_replica": 1082752,
    "kv_dtype": "turboquant_4bit_nc",
    "engine": "clavue-2b-v61",
    "stable_alias": "clavue-2b-last",
    "page": "route-2026-09-24.html"
  },
  "serving_27b": {
    "deploy": "vllm-qwen38-27b-lora",
    "tensor_parallel": 2,
    "replicas": 10,
    "replicas_core": 10,
    "replicas_elastic_planned": {"ai-022": 3, "ai-032": 1},
    "max_model_len": 131072,
    "max_num_seqs_per_replica": 64,
    "full_context_concurrency_per_replica": 7.41,
    "kv_cache_tokens_per_replica": 971124,
    "kv_dtype": "turboquant_3bit_nc",
    "page": "llm-27b.html"
  },
  "base_url": "https://ttkk.inping.com/v1",
  "intranet_docs": "http://192.168.2.100:31803/",
  "tech_center": "http://192.168.2.100:8099/tech-center/",
  "intranet_27b": "http://192.168.2.100:11401/v1",
  "intranet_35b": "http://192.168.2.100:30081/v1",
  "auth": "Authorization: Bearer sk-…",
  "docs": "https://ttkk.inping.com/docs/",
  "models": [
    {"id": "clavue-2.1", "cluster": "tools-clean", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm.html", "title": "对话 · 质量"},
    {"id": "qwen3.8:27b", "cluster": "qwen3.8:27b", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm-27b.html", "title": "27B 裸引擎（零注入）"},
    {"id": "tools-clean", "cluster": "tools-clean", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm.html", "title": "27B QLoRA tools-clean"},
    {"id": "clavue-2-code", "cluster": "tools-clean", "pipeline": "llm-code-retrieve", "endpoint": "/v1/chat/completions", "page": "retrieve-front.html", "title": "27B 编码 + 仓库检索前置"},
    {"id": "clavue-2.1-fast", "cluster": "clavue-2b-last", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "route-2026-09-24.html", "title": "对话 · 速度（稳定名 clavue-2b-last，当前引擎 clavue-2b-v61）"},
    {"id": "qwen3.6:35b-a3b", "cluster": "qwen3.6:35b-a3b", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm-35b.html", "title": "35B 裸引擎（零注入）"},
    {"id": "lingsi1.0", "cluster": "qwen3.5:4b", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm.html", "title": "多模态小模型"},
    {"id": "auto", "cluster": "qwen3.5:4b", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm.html", "title": "面板默认"},
    {"id": "luxtts", "cluster": "luxtts", "pipeline": "tts-live", "endpoint": "/v1/audio/speech", "page": "tts-live.html", "title": "实时短句语音，默认夹子", "aliases": ["clavue-tts-live"]},
    {"id": "clavue-tts-ml", "cluster": "qwen3-tts-acted", "pipeline": "tts", "endpoint": "/v1/audio/speech", "page": "tts.html", "title": "有声小说男主组", "aliases": ["qwen3-tts-acted-ml"]},
    {"id": "clavue-tts-fn", "cluster": "qwen3-tts-acted", "pipeline": "tts", "endpoint": "/v1/audio/speech", "page": "tts.html", "title": "有声小说女旁白", "aliases": ["qwen3-tts-acted-fn"]},
    {"id": "clavue-tts-rest", "cluster": "qwen3-tts-acted", "pipeline": "tts", "endpoint": "/v1/audio/speech", "page": "tts.html", "title": "有声小说其余声线", "aliases": ["qwen3-tts-acted-rest"]},
    {"id": "clavue-tts", "cluster": "qwen3-tts-acted", "pipeline": "tts", "endpoint": "/v1/audio/speech", "page": "tts.html", "title": "有声小说兼容名（八路混批）", "aliases": ["clavue-tts-acted", "qwen3-tts-acted"]},
    {"id": "qwen3-tts-asmr", "cluster": "qwen3-tts-asmr", "pipeline": "tts-asmr", "endpoint": "/v1/audio/speech", "page": "asmr.html", "title": "耳语 ASMR（慢服务 20-110s/句，中文 aiden/vivian 最佳）", "aliases": ["clavue-asmr"]},
    {"id": "clavue-img", "cluster": "z-image-turbo", "pipeline": "text-to-image", "endpoint": "/v1/images/generations", "page": "image.html", "title": "文生图（默认 WebP，output_format=jpeg|png）"},
    {"id": "z-image-turbo", "cluster": "z-image-turbo", "pipeline": "text-to-image", "endpoint": "/v1/images/generations", "page": "image.html", "title": "文生图引擎名（默认 WebP，output_format=jpeg|png）"},
    {"id": "qwen-image-21", "cluster": "qwen-image-21", "pipeline": "image-edit", "endpoint": "/v1/images/edits", "page": "image.html", "title": "指令改图（Qwen-Image-2.1 · 唯一 model id）"},
    {"id": "clavue-seg-v1", "cluster": "clavue-seg-v1", "pipeline": "image-seg", "endpoint": "/v1/images/segments", "page": "image.html", "title": "视觉分段"},
    {"id": "clavue-inpaint-v1", "cluster": "clavue-inpaint-v1", "pipeline": "image-edit", "endpoint": "/v1/images/edits", "page": "image.html", "title": "区域编辑"},
    {"id": "clavue-tryon-v1", "cluster": "clavue-tryon-v1", "pipeline": "image-tryon", "endpoint": "/v1/images/try-on", "page": "image.html", "title": "虚拟换衣"},
    {"id": "clavue-music", "cluster": "minimax-music3", "pipeline": "music", "endpoint": "/v1/audio/speech", "page": "music.html", "title": "音乐生成（单次最长 260 秒，默认 MP3；时长用 audio_duration 与 voice 同时传）", "aliases": ["minimax-music3"]},
    {"id": "funasr", "cluster": "sensevoice", "pipeline": "asr", "endpoint": "/v1/audio/transcriptions", "page": "asr.html", "title": "语音识别旧名。现为 SenseVoice，内网 11401"},
    {"id": "faster-whisper-medium", "cluster": "sensevoice", "pipeline": "asr", "endpoint": "/v1/audio/transcriptions", "page": "asr.html", "title": "语音识别旧名。请求转到 SenseVoice，不启动 whisper 副本"},
    {"id": "sensevoice", "cluster": "sensevoice", "pipeline": "asr", "endpoint": "/v1/audio/transcriptions", "page": "asr.html", "title": "SenseVoice-Small 语音识别。与 funasr、faster-whisper-medium 同一引擎"},
    {"id": "clavue-jev", "cluster": "jev-omni-zh", "pipeline": "jev-judge", "endpoint": "POST /v1/judge · ttkk/ttqq/api/11401", "page": "jev.html", "title": "JEV 多选判官", "aliases": ["clavue-jev"], "note": "非 chat completions"},
    {"id": "clavue-ocr", "cluster": "PaddleOCR-VL-1.6-0.9B", "pipeline": "ocr", "endpoint": "/v1/chat/completions", "page": "ocr.html", "title": "OCR", "aliases": ["PaddleOCR-VL-1.6-0.9B", "PaddleOCR-VL-1.5-0.9B"]},
    {"id": "clavue-emb", "cluster": "bge-m3", "pipeline": "embedding", "endpoint": "/v1/embeddings", "page": "ocr.html", "title": "文本嵌入", "aliases": ["bge-m3"]},
    {"id": "clavue-trans", "cluster": "clavue-trans", "pipeline": "translation", "endpoint": "/v1/trans/translate", "page": "trans.html", "title": "OCR+翻译+Jev", "aliases": ["clavue-translate"]},
    {"id": "clavue-strata-125b", "cluster": "clavue-strata-125b", "pipeline": "llm-chat", "endpoint": "/v1/chat/completions", "page": "llm.html", "title": "Strata 125B MoE（128K 单线 · 029×4 + 2080Ti 004/005 · IQ2_XS）", "aliases": ["clavue-strata-125b-128k", "strata-iq2-xs", "qwen3.8-flash-next-125b", "qwen3.8-flash-next-iq2_xs"]}
  ]
}