Thinking Machines develops AI that can listen and talk simultaneously
Original: Thinking Machines wants to build an AI that actually listens while it talks
Why This Matters
Represents potential breakthrough in AI conversational interfaces mimicking natural human interaction patterns
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, announced interaction models that enable AI to process input and generate responses simultaneously. The TML-Interaction-Small model responds in 0.40 seconds, matching human conversation speed.
Thinking Machines Lab unveiled interaction models that break from traditional AI conversation patterns where users and AI take turns speaking. The startup's TML-Interaction-Small model uses full duplex technology to process input while generating responses simultaneously, achieving 0.40-second response times comparable to natural human conversation and faster than OpenAI and Google models. Founded by former OpenAI CTO Mira Murati, the company positions this as native interactivity rather than an add-on feature. The technology remains in research preview with limited testing planned for coming months and wider release later this year.