{
  "contentStorage": "postgres",
  "title": "Gemini 3.8 Live Extended Thinking",
  "description": "Google DeepMind 于 2026 年 9 月 15 日 发布的 Gemini 3.8 Live Extended Thinking。",
  "url": "/ms/works/gemini-3-8-live-extended-thinking/at/1.html",
  "entity": {
    "id": "work_gemini_3_8_live_extended_thinking",
    "type": "work",
    "title": "Gemini 3.8 Live Extended Thinking",
    "slug": "gemini-3-8-live-extended-thinking",
    "summary": "Google DeepMind 于 2026 年 9 月 15 日 发布的 Gemini 3.8 Live Extended Thinking。",
    "status": "published",
    "categories": [
      "category-science-technology"
    ],
    "tags": [],
    "sources": [
      {
        "id": "s1",
        "title": "Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking",
        "url": "https://deepmind.google/blog/introducing-gemini-3-8-live-and-3-8-live-extended-thinking/"
      },
      {
        "id": "s2",
        "title": "Gemini 3.8 Live Extended Thinking",
        "url": "https://aistudio.google.com/docs/models/gemini-3.8-live-extended-thinking"
      },
      {
        "id": "s3",
        "title": "Gemini Developer API pricing",
        "url": "https://ai.google.dev/gemini-api/docs/pricing"
      },
      {
        "id": "s4",
        "title": "Thinking in the Live API",
        "url": "https://aistudio.google.com/docs/live-api/thinking"
      }
    ],
    "createdAt": "2026-11-11T13:45:49+08:00",
    "updatedAt": "2026-11-11T13:45:49+08:00",
    "updatedBy": "user:admin",
    "revision": 1,
    "body": "## 发布背景\n\nGemini 3.8 Live Extended Thinking 是 [[organization_google_deepmind|Google DeepMind]] 于 2026 年 9 月 15 日 发布的 Gemini 模型[^s1]。它是 Live API 的高推理音频到音频模型，可在说话的同时进行后台推理并播报多步任务进度。[^s1]\n\n## 定价与获取\n\n每百万 token 输入文本 0.75 美元、音频 3.00 美元、图像视频 1.00 美元，输出文本 4.50 美元、音频 12.00 美元，思考 token 计入输出计费[^s3]。API 标识为 gemini-3.8-live-extended-thinking[^s1]。\n\n## 性能表现\n\n以下均为 Google DeepMind 在发布公告与技术报告中公布的厂商自报成绩，尚未经独立复现：\n\n- Artificial Analysis Speech to Speech Quality Index：82.6，厂商自报，位列第一[^s1]。\n- τ-Voice：68.6%，厂商自报，语音智能体任务完成率[^s1]。\n- Sierra τ-Voice-banking：35.1%，厂商自报，银行场景语音任务完成率[^s1]。\n- Big Bench Audio：97.7%，厂商自报，音频推理得分[^s1]。\n\n## 主要特点\n\n上下文窗口 131,072 token，最大输出 65,536 token[^s1]。\n\n使用时必须指定 thinking_level（low、medium 或 high），且仅支持异步函数调用，不支持同步阻塞工具[^s1]。\n\n它与 [[work_gemini_3_8_live|Gemini 3.8 Live]] 同日发布并共享同一价目表，定位为面向复杂多步任务的高推理版本[^s1]。\n\n## 定位\n\n它是 Gemini 3.8 Live 的高推理同系模型，接替 Gemini Live 2.5 Flash 原生音频系列面向复杂多步实时语音智能体，尚无已公布的后继型号。截至 2026 年 11 月，该模型仍在提供。",
    "url": "/works/gemini-3-8-live-extended-thinking.html",
    "lang": "ms",
    "sourceLang": "zh",
    "fallback": true,
    "availableLangs": [
      "zh"
    ],
    "workKind": "ai_model"
  },
  "entities": null,
  "relations": [
    {
      "id": "relation_work_gemini_3_8_live_extended_thinking_published_by_organization_google_deepmind",
      "fromId": "work_gemini_3_8_live_extended_thinking",
      "relationType": "published_by",
      "toId": "organization_google_deepmind",
      "sourceRefs": [
        {
          "entityId": "work_gemini_3_8_live_extended_thinking",
          "sourceId": "s1"
        }
      ],
      "label": "Published by",
      "targetId": "organization_google_deepmind",
      "targetTitle": "Google DeepMind",
      "targetUrl": "/ms/organizations/google-deepmind.html",
      "incoming": false
    }
  ],
  "versions": [
    {
      "revision": 1,
      "updatedAt": "2026-11-11T13:45:49+08:00",
      "message": "当前版本",
      "hasSnapshot": true,
      "url": "/ms/works/gemini-3-8-live-extended-thinking/at/1.html"
    }
  ],
  "viewRevision": 1,
  "lang": "ms",
  "availableLangs": [
    "zh"
  ],
  "fallback": true,
  "infoFields": [
    {
      "key": "work_kind",
      "label": "Work kind",
      "value": "Model AI"
    }
  ],
  "relationsTotal": 1,
  "relationCounts": [
    {
      "relationType": "published_by",
      "incoming": false,
      "label": "Published by",
      "total": 1
    }
  ]
}