Boson AI 发布 45B 全双工模型 Higgs Realtime Duplex
Boson AI 称 Higgs Realtime Duplex 是一款 45B 混合专家模型,可同时听、看、说,并在说话时于后台进行思考、调用工具和委派更复杂任务。厂商介绍其采用 160 毫秒分块方案处理音频输入输出、周期性视频帧与人脸输出潜变量,并通过多教师在策略蒸馏与模拟环境自我改进循环训练。该发布不提供开放权重,Duplex API 处于等候名单,之前的轮次式 Higgs Realtime 已可用;Boson AI 承认七项基准中有五项由 OpenAI 最佳设置领先,全双工基准差距最大。
该帖详细披露了在同一数据流中并行处理音视频文本与工具调用、并在后台委派任务的一手设计思路,可与轮次式语音模型及 Thinking Machines 的后台模型路线对照。
Disclosure: I'm at Boson AI and this is our release. Not open weights; the Duplex API has a waitlist, and the previous turn-based Higgs Realtime is available now.
Most voice models are still turn-based: listen, wait for end of turn, think, speak. Duplex runs all streams at once and keeps working in the background while it talks.
Streams. Everything is cut into 160 ms chunks:
- Audio in: 1 token per chunk (6.25 tokens/s)
- Audio out: 2 tokens per chunk (12.5 tokens/s)
- Video in: one ~12-token camera frame every 6 chunks
- Face out: 2 latents per chunk
- Text in and out alongside
Silence is represented as chunks too, which is what the model uses to decide when to backchannel, interrupt or stay quiet. That comes to ~2.6K tokens per minute for audio and video, 2.6K to 5K with thinking and tool traces (progressively compacted). The last 5 minutes plus compacted text history is 13K to 25K tokens; older history is kept as text.
#Boson AI#Higgs Realtime Duplex#Higgs Realtime#TML Interaction Small#Thinking Machines#OpenAI#Google