Thinking Machines shows a model that listens, thinks, and speaks at once
- Source
- x.com
- Date
People talk, listen, watch, think, and collaborate at the same time, in real time. We've designed an AI that works with people the same way. We share our approach, early results, and a quick look at our model in action. https://t.co/AFJZ5kH7Ku
While Lilian is telling a story, the interaction model can track when she is thinking, yielding, self-correcting, or inviting a response; there is no specific built dialogue management system.
Context
Thinking Machines' interaction model treats conversation as continuous input and output, so it can keep listening while speaking. Its technical account describes interleaving 200-millisecond chunks of incoming and outgoing information. Silence, interruptions and overlapping speech remain part of the model's , rather than requiring every exchange to wait for a completed user turn. Architecture.
Longer reasoning and tool tasks go to a background model that shares the conversation context. The interaction model continues responding and incorporates results as they arrive. This division is intended to preserve conversational responsiveness while more demanding work proceeds, explaining how an assistant could keep talking during a search. System overview.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
Checking sign-in…
Loading comments…