chore: update bailian-docs-llm-wiki (2026-09-23)

Author: gujieye

skills/bailian-docs-llm-wiki/llms.txt

@@ -114,6 +114,7 @@
 {"slug":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","description":"Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","contextWindow":53248,"capabilities":["Realtime-ASR"]}],"detailPath":"groups/qwen3-livetranslate-flash.json","maxContextWindow":53248}
 {"slug":"qwen3-max","name":"Qwen3-Max","description":"千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3-max","name":"Qwen3-Max","contextWindow":262144,"capabilities":["TG","Reasoning"]},{"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","contextWindow":262144,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3-max.json","maxContextWindow":262144}
 {"slug":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","description":"千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","contextWindow":65536,"capabilities":["ASR"]}],"detailPath":"groups/qwen3-omni-30b-a3b-captioner.json","maxContextWindow":65536}
+{"slug":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","description":"Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","contextWindow":65536,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3-omni-flash-realtime.json","maxContextWindow":65536}
 {"slug":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","description":"Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","contextWindow":65536,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3-omni-flash.json","maxContextWindow":65536}
 {"slug":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","description":"Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash-realtime.json"}
 {"slug":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash.json"}
@@ -155,7 +156,6 @@
 {"slug":"speech-biasing","name":"语音识别热词","description":"热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"speech-biasing","name":"语音识别热词","capabilities":["ASR"]}],"detailPath":"groups/speech-biasing.json"}
 {"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["stepfun"],"itemCount":2,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG","VU"]},{"model":"stepfun/step-5-preview","name":"stepfun/step-5-preview","contextWindow":1048576,"capabilities":["TG","VU","Reasoning"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":1048576}
 {"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192}
-{"slug":"tongyi-xiaomi-analysis-flash","name":"伶鹊-对话分析-flash","description":"伶鹊-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"伶鹊-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768}
 {"slug":"tongyi-xiaomi-analysis-pro","name":"伶鹊-对话分析-pro","description":"伶鹊-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"伶鹊-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768}
 {"slug":"tripo-models-market-place","name":"Tripo","description":"AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。","primaryCapability":"3D-generation","capabilities":["3D-generation"],"providers":["tripo"],"itemCount":2,"items":[{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","capabilities":["3D-generation"]},{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","capabilities":["3D-generation"]}],"detailPath":"groups/tripo-models-market-place.json"}
 {"slug":"unisound-models-market-place","name":"云知声文本模型","description":"由云知声提供的 Unisound U2 模型 API 服务.","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["unisound"],"itemCount":1,"items":[{"model":"unisound/unisound-u2","name":"unisound/unisound-u2","contextWindow":163840,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/unisound-models-market-place.json","maxContextWindow":163840}

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

@@ -60,8 +60,8 @@
       "samples": {
         "dashscope": {
           "default": {
-            "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n     --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n     --header \"Content-Type: application/json\" \\\n     --header \"X-DashScope-SSE: disable\" \\\n     --data '{\n    \"model\": \"qwen-audio-3.1-asr-flash-message\",\n    \"input\": {\n        \"messages\": [\n            {\n                \"role\": \"user\",\n                \"content\": [\n                    {\n                        \"type\": \"input_audio\",\n                        \"input_audio\": {\n                            \"data\": \"{YOUR_AUDIO_URL}\"\n                        }\n                    }\n                ]\n            }\n        ]\n    },\n    \"parameters\": {\n        \"format\": \"wav\",\n        \"sample_rate\": \"16000\"\n    }\n}'",
-            "docUrl": "https://docs.bailian.console.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide"
+            "python": "from http import HTTPStatus\nimport dashscope\nfrom dashscope.audio.asr import Recognition\nimport os\n\n\ndashscope.api_key = os.environ.get('DASHSCOPE_API_KEY')\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nrecognition = Recognition(model='qwen-audio-3.1-asr-flash-message',\n                          format='wav',\n                          sample_rate=16000,\n                          callback=None)\nresult = recognition.call('{YOUR_AUDIO_FILE}')\nif result.status_code == HTTPStatus.OK:\n    print('Recognition result: ')\n    print(result.get_sentence())\nelse:\n    print('Error: ', result.message)\n    \nprint(\n    '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n    .format(\n        recognition.get_last_request_id(),\n        recognition.get_first_package_delay(),\n        recognition.get_last_package_delay(),\n    ))",
+            "docUrl": "https://docs.bailian.console.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide"
           }
         }
       }

skills/bailian-docs-llm-wiki/models/groups/qwen-audio-3.1-asr-message.json

@@ -18,7 +18,8 @@
       "description": "Qwen3.8-Omni-Flash-Realtime 通过多种实时协议提供音视频双工交互，面向智能硬件、机器人与实时交互 Agent。支持多通道音频输入与多种通道布局，视频表征可在细粒度与聚合之间按精度与开销选择，支持 60+ 语言音频输入和 30+ 语言语音输出；工具侧支持 Function Call 与 MCP 服务接入，具备工具发现、调用审批与结果续轮的完整事件链。可以在高速实时响应的同时，编排复杂业务逻辑和支持上百个工具的调用，同时拥有生动、多样的声音生成能力。",
       "collectionTag": "qwen3.8",
       "features": [
-        "web-search"
+        "web-search",
+        "function-calling"
       ],
       "provider": "qwen",
       "model": "qwen3.8-omni-flash-realtime",