Thinking Machines, founded by ex-OpenAI CTO Mira Murati, is demonstrating 'interaction models' that respond to users in real time across audio, video, and text.

Thinking Machines announced Monday that it is working on 'interaction models' designed to let people collaborate with AI by continuously taking in audio, video, and text, and responding in real time.

Unlike current models that wait for users to finish typing or speaking before perceiving input, interaction models are intended to process multiple modalities simultaneously, addressing what Thinking Machines describes as a 'bandwidth bottleneck' in human-AI collaboration.

Thinking Machines plans to open a 'limited research preview' in the 'coming months' and aims to do a 'wider release later this year'; the company was founded by Murati in February 2025 after she left OpenAI.