chore: update bailian-docs-llm-wiki (2026-07-15)

skills/bailian-docs-llm-wiki/SKILL.md

@@ -104,6 +104,7 @@ description: >-
 | `ME` | 多模态嵌入 |
 | `TR` | 翻译 |
 | `3D-generation` | 3D 生成 |
+| `Realtime-Chatting` | Realtime-Chatting |
 
 一个模型常常带多个 capability，`index.md` 中按 `capabilities[0]`（主能力）归类，
 查找时按中文标签即可定位章节。
@@ -147,22 +148,22 @@ description: >-
 | **按家族筛选**：按 primaryCapability / providers / itemCount / maxContextWindow 找家族 | `models/families.jsonl`（一行一家族，含 items[] 摘要） |
 | 模型家族总览 / 按能力分桶浏览         | `models/index.md`                                        |
 | 主题页 / API 文档（按功能领域查找）    | `wiki/index.md`（完整索引入口）                          |
-| 函数调用（Function Calling / 工具调用） | `wiki/concepts/function-calling.md` |
+| 函数调用（Function Calling） | `wiki/concepts/function-calling.md` |
 | 检索增强生成（RAG） | `wiki/concepts/rag.md` |
+| Token 与计量计费 | `wiki/concepts/token.md` |
+| 流式输出 | `wiki/concepts/streaming.md` |
 | OpenAI 兼容接口 | `wiki/concepts/openai-compatible-interface.md` |
 | API Key 鉴权 | `wiki/concepts/api-key.md` |
 | 异步调用与任务轮询 | `wiki/concepts/async-invocation.md` |
-| Token 与计费 | `wiki/concepts/token-billing.md` |
 | 业务空间（Workspace） | `wiki/concepts/workspace.md` |
-| 流式输出 | `wiki/concepts/streaming.md` |
-| DashScope SDK | `wiki/concepts/dashscope-sdk.md` |
-| 多模态 | `wiki/concepts/multimodal.md` |
-| 模型微调与模型压缩对比 | `wiki/comparisons/fine-tuning-vs-model-compression.md` |
-| 模型部署与高速推理对比 | `wiki/comparisons/model-deployment-vs-high-speed-inference.md` |
-| 知识库与记忆库对比 | `wiki/comparisons/knowledge-base-vs-memory-library.md` |
-| 图像、视频与 3D 生成对比 | `wiki/comparisons/image-vs-video-vs-3d-generation.md` |
-| 托管智能体与 LLM 应用对比 | `wiki/comparisons/managed-agents-vs-llm-application.md` |
-| 模型评估与应用评估对比 | `wiki/comparisons/model-evaluation-vs-application-evaluation.md` |
+| 模型微调与生产链路 | `wiki/concepts/fine-tuning.md` |
+| 评测体系 | `wiki/concepts/evaluation.md` |
+| 多模态能力 | `wiki/concepts/multimodal.md` |
+| 图像、视频与3D生成对比 | `wiki/comparisons/media-generation-compare.md` |
+| 模型微调、压缩与高速推理对比 | `wiki/comparisons/model-optimization-compare.md` |
+| 知识库与长期记忆对比 | `wiki/comparisons/knowledge-memory-compare.md` |
+| 应用监控与模型监控对比 | `wiki/comparisons/monitoring-compare.md` |
+| 应用评测与模型评测对比 | `wiki/comparisons/evaluation-compare.md` |
 
 > 实际文件名以 `wiki/index.md` 为准；上表若有出入应回到索引页查找。
 

skills/bailian-docs-llm-wiki/models/families.jsonl

@@ -53,6 +53,9 @@
 {"slug":"pixverse-v6-market-place","name":"PixVerse V6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-v6-market-place.json"}
 {"slug":"qvq-max","name":"QVQ-Max","description":"千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-max","name":"QVQ-Max","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-max.json","maxContextWindow":131072}
 {"slug":"qvq-plus","name":"Qwen-QVQ-Plus","description":"千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-plus","name":"QVQ-Plus","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-plus.json","maxContextWindow":131072}
+{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":8192}
+{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":8192}
+{"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"}
 {"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072}
 {"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072}
 {"slug":"qwen-deep-research","name":"qwen-deep-research","description":"千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-deep-research","name":"qwen-deep-research","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-deep-research.json","maxContextWindow":1000000}
@@ -142,7 +145,8 @@
 {"slug":"vanchin-models-market-place","name":"Vanchin DeepSeek","description":"由快手万擎提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["deepseek"],"itemCount":6,"items":[{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","contextWindow":8192,"capabilities":["VU","TG"]},{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","contextWindow":131072,"capabilities":["TG"]},{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/vanchin-models-market-place.json","maxContextWindow":1048576}
 {"slug":"video-style-transform","name":"视频风格重绘","description":"视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"video-style-transform","name":"视频风格重绘","capabilities":["VG"]}],"detailPath":"groups/video-style-transform.json"}
 {"slug":"videoretalk","name":"声动人像VideoRetalk","description":"VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"videoretalk","name":"声动人像VideoRetalk","capabilities":["VG"]}],"detailPath":"groups/videoretalk.json"}
-{"slug":"vidu-models-market-place","name":"Vidu AI生视频","description":"由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。","primaryCapability":"VG","capabilities":["VG"],"providers":["vidu"],"itemCount":17,"items":[{"model":"vidu/viduq2_reference2video","name":"ViduQ2_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2_text2video","name":"ViduQ2_text2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_img2video","name":"ViduQ2-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_reference2video","name":"ViduQ2-Pro_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_start-end2video","name":"ViduQ2-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro-fast_img2video","name":"ViduQ2-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_img2video","name":"ViduQ2-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_start-end2video","name":"ViduQ2-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3_reference2video","name":"ViduQ3_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-mix_reference2video","name":"ViduQ3-mix_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_img2video","name":"ViduQ3-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_start-end2video","name":"ViduQ3-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_text2video","name":"ViduQ3-Pro_text2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_img2video","name":"ViduQ3-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_reference2video","name":"ViduQ3-Turbo_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_start-end2video","name":"ViduQ3-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","capabilities":["VG"]}],"detailPath":"groups/vidu-models-market-place.json"}
+{"slug":"vidu-image-models-market-place","name":"Vidu AI生图","description":"由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。","primaryCapability":"IG","capabilities":["IG"],"providers":["vidu"],"itemCount":4,"items":[{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","capabilities":["IG"]},{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","capabilities":["IG"]}],"detailPath":"groups/vidu-image-models-market-place.json"}
+{"slug":"vidu-models-market-place","name":"Vidu AI生视频","description":"由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。","primaryCapability":"VG","capabilities":["VG"],"providers":["vidu"],"itemCount":20,"items":[{"model":"vidu/viduq2_reference2video","name":"ViduQ2_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2_text2video","name":"ViduQ2_text2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_img2video","name":"ViduQ2-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_reference2video","name":"ViduQ2-Pro_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_start-end2video","name":"ViduQ2-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro-fast_img2video","name":"ViduQ2-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_img2video","name":"ViduQ2-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_start-end2video","name":"ViduQ2-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3_reference2video","name":"ViduQ3_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-ad_reference2video","name":"ViduQ3-Ad_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-drama_reference2video","name":"ViduQ3-Drama_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-mix_reference2video","name":"ViduQ3-mix_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_img2video","name":"ViduQ3-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_start-end2video","name":"ViduQ3-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_text2video","name":"ViduQ3-Pro_text2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro-fast_img2video","name":"ViduQ3-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_img2video","name":"ViduQ3-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_reference2video","name":"ViduQ3-Turbo_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_start-end2video","name":"ViduQ3-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","capabilities":["VG"]}],"detailPath":"groups/vidu-models-market-place.json"}
 {"slug":"virtualmodel-v2","name":"虚拟模特V2","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"virtualmodel-v2","name":"虚拟模特V2","capabilities":["IG"]}],"detailPath":"groups/virtualmodel-v2.json"}
 {"slug":"voice-enrollment","name":"大模型声音复刻及声音设计","description":"大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","capabilities":["TTS"]}],"detailPath":"groups/voice-enrollment.json"}
 {"slug":"wan-image-edit","name":"Wan-Image","description":"指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":5,"items":[{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","capabilities":["IG"]},{"model":"wan2.6-image","name":"Wan2.6-Image","capabilities":["IG"]},{"model":"wan2.7-image","name":"Wan2.7-Image","capabilities":["IG"]},{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","capabilities":["IG"]},{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","capabilities":["IG"]}],"detailPath":"groups/wan-image-edit.json"}