[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fNf4GmYUeOv0Tni5qNbe9BPfBiaXCicvVa3XtO1poRUY":3,"$fdoG5HuuyISXeZniATKtBK8B7wTFlvSx4PY_50DKMn70":37},{"id":4,"name":5,"code":6,"logo":7,"cover":8,"url":9,"categoryId":10,"categoryName":11,"vendorId":12,"vendorName":13,"intro":14,"features":15,"pros":16,"cons":17,"pricingType":18,"priceNote":19,"freeQuota":20,"platform":21,"supportApi":22,"apiUrl":23,"score":24,"isRecommend":25,"isHot":25,"click":25,"sort":25,"createTime":26,"tags":27,"content":36},206,"MOSS-Live","moss-live","\u002Flogos\u002Fmoss.svg",null,"https:\u002F\u002Fwww.moss.cn",7,"AI音频",142,"模思智能","模思智能发布的全双工实时交互系统，边说边听、语义判停，支持视频画面与屏幕共享理解","语义级全双工：边说边听，动态判断回应、等待、接受打断或保持沉默;持续多模态感知：语音流、视频画面、声学环境三路融合;前后台协同：前台保实时对话，后台异步跑复杂推理与长程任务;支持实时视频通话与屏幕共享理解，能看着你说的东西对话;流式同声传译，对话中可随时调用工具;开放 API，可嵌入自有产品，当前限时免费","把节奏做成了可量化的技术指标（停顿抢话率 1.44%），解决的是真实对话体验;前后台分离兼顾低延迟与复杂任务，不必二选一;已开放 API 且限时免费，试错成本几乎为零","是交互系统与 API，不是开箱即用的 App，普通用户只能体验 Demo;官方对比数据来自自建评测，第三方复现结果尚未见到;限时免费之后的定价策略未公布，长期成本不确定",2,"API 限时免费开放，后续计费政策以 platform.moss.cn 为准","限时免费体验与 API 接入","Web\u002FAPI",1,"https:\u002F\u002Fplatform.moss.cn",8.4,0,"2026-11-06 16:16:11",[28,32],{"id":29,"name":30,"color":31,"type":22},8,"国产自研","#F56C6C",{"id":33,"name":34,"color":35,"type":22},5,"支持API","#00B4A0","MOSS-Live 是上海模思智能科技在 2026 年 11 月发布的全双工实时交互系统，要解决的是语音 AI 长期以来的一个尴尬：它遵循近似半双工的轮流发言模式，你说完它再说，听和说很难真正并行。于是就有了那些熟悉的瞬间：你话没说完它已经抢答，你随口嗯一声它以为你要打断，聊到新话题再让它接着刚才的说，它已经忘了讲到哪。MOSS-Live 的做法是让 AI 在说话的同时持续倾听，实时理解语义、说话意图与对话状态，动态判断何时回应、何时等待、何时接受打断、何时保持沉默。它在持续多模态感知上同时融合三路信息：语音流、视频画面与声学环境——比如结合实时画面识别你手指的物体，或识别导航播报这类无关声音并主动不响应。架构上采用前后台协同：前台交互模型负责持续感知、理解与语音交流，保证低延迟；遇到深度推理或长程任务，就把上下文交给后台模型异步执行，前台继续陪用户聊，结果回来再主动反馈，两者通过记忆、上下文与任务状态管理协同。它还支持实时视频通话、屏幕共享理解、实时调用工具与流式同声传译。官方公布的对比评测里，FDB v1 停顿抢话率（不含附和）约 1.44%，明显低于对照的 GPT Realtime 2.1 与豆包 SeedDuplex，无关语音误接话等指标同样领先。目前 MOSS-Live 已开放 API 接入并限时免费，可在模思智能官网体验，在 platform.moss.cn 申请密钥。母公司此前在 2026 年 1 月发布过语音大模型 MOSS-Speech，并参与发布了开源音视频同步生成模型 MOVA。",[38,42,44,46,50,54,59,64,68,72,76],{"id":18,"name":39,"color":40,"type":22,"toolCount":41},"免费增值","#409EFF",114,{"id":29,"name":30,"color":31,"type":22,"toolCount":43},69,{"id":33,"name":34,"color":35,"type":22,"toolCount":45},37,{"id":22,"name":47,"color":48,"type":22,"toolCount":49},"完全免费","#67C23A",32,{"id":10,"name":51,"color":52,"type":22,"toolCount":53},"移动端可用","#FF8A65",31,{"id":55,"name":56,"color":57,"type":22,"toolCount":58},4,"开源","#909399",15,{"id":60,"name":61,"color":62,"type":22,"toolCount":63},3,"付费使用","#E6A23C",12,{"id":65,"name":66,"color":67,"type":22,"toolCount":29},6,"可私有化","#7B61FF",{"id":69,"name":70,"color":71,"type":22,"toolCount":18},10,"长上下文","#3B55D9",{"id":73,"name":74,"color":75,"type":22,"toolCount":18},11,"多模态理解","#9C27B0",{"id":77,"name":78,"color":79,"type":18,"toolCount":49},16,"智能体","#10b981"]