Thinking Machines Unveils Real-Time Interaction Model for Humanlike AI
Thinking Machines Lab Inc., an AI startup founded by former OpenAI CTO Mira Murati, has announced a research preview of its new 'interaction models.' These multimodal systems aim to eliminate the latency inherent in traditional turn-based AI interactions by enabling full-duplex communication. The new architecture allows the AI to listen, see, and speak simultaneously, processing inputs and outputs in 200-millisecond chunks. The system features a dual-model design: TML-Interaction-Small, a 276-billion parameter model handling immediate dialogue, and an asynchronous background agent for complex reasoning and tool use. By utilizing encoder-free early fusion, the model achieves a turn-taking latency of under 0.4 seconds, outperforming competitors like Google’s Gemini and OpenAI’s GPT-realtime. This advancement promises to transform enterprise applications, such as real-time safety monitoring in manufacturing and more natural customer service interactions, by allowing AI to react to visual and auditory cues without the delays that currently force users to adapt their communication style.
Wire timeline
Thinking Machines Unveils Real-Time Interaction Model for Humanlike AI
Thinking Machines Lab Inc., an AI startup founded by former OpenAI CTO Mira Murati, has announced a research preview of its new 'interaction models.' These multimodal systems aim to eliminate the latency inherent in traditional turn-based AI interactions by enabling full-duplex communication. The new architecture allows the AI to listen, see, and speak simultaneously, processing inputs and outputs in 200-millisecond chunks. The system features a dual-model design: TML-Interaction-Small, a 276-billion parameter model handling immediate dialogue, and an asynchronous background agent for complex reasoning and tool use. By utilizing encoder-free early fusion, the model achieves a turn-taking latency of under 0.4 seconds, outperforming competitors like Google’s Gemini and OpenAI’s GPT-realtime. This advancement promises to transform enterprise applications, such as real-time safety monitoring in manufacturing and more natural customer service interactions, by allowing AI to react to visual and auditory cues without the delays that currently force users to adapt their communication style.
SiliconANGLE