Thinking Machines Lab, an artificial intelligence startup founded by Mira Murati, has introduced a groundbreaking category of AI systems known as 'interaction models.' These models are designed to facilitate real-time communication, addressing a major limitation of conventional AI systems that often create awkward pauses during conversations. Unlike traditional models that wait for users to finish speaking or typing, the new architecture allows for continuous listening, processing, and responding. This 'full-duplex' interaction enables the AI to break communication into micro-turns of approximately 200 milliseconds, significantly enhancing the fluidity of conversations. The centerpiece of this system is the TML-Interaction-Small model, which features 276 billion parameters and achieves a response latency of under 0.4 seconds. This performance surpasses competitors like Google’s Gemini-3.1-flash-live and GPT-realtime-2.0. The implications of this technology extend beyond customer service, potentially transforming industries such as healthcare and manufacturing by enabling real-time monitoring and immediate responses to critical situations. Currently, these interaction models are available to a limited number of research partners, with a broader release planned for later this year.