Updated

Higgs Realtime Duplex (announced)

Boson AI introduced Higgs Realtime Duplex, a full-duplex interactive multimodal model that listens, watches, and talks while running background reasoning and tool use. Its announcement describes 160 ms interaction chunks and a 45B mixture-of-experts front model; public model access is not documented yet.

By No Code MBA · Facts checked

Provider
Boson AI
Announcement date
Status at last check
Announced

What changed

Boson AI describes Higgs Realtime Duplex as the next generation of Higgs Realtime and its first full-duplex interactive omni model. The company says it can listen, watch, talk, render a face, and run background thinking, tool calls, and delegation simultaneously, using 160 ms chunks. Boson identifies the front model as a 45B mixture of experts trained on about 2 trillion tokens of text, audio, and video.

Access and availability

Boson AI has announced the model and published demonstrations. The announcement says an API is coming soon; it does not document general access or a downloadable model.

Official access documentation

What you can build

When access opens, evaluate its interruption handling and background task results on a bounded, human-reviewed workflow before relying on it.

Limitations

The release is an announcement rather than a documented general-availability launch. No public API documentation, model download, or access requirements are provided in the announcement. Capability and technical details are from Boson AI.

Sources and release history