Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, has announced 'interaction models' — AI that can process input and generate responses simultaneously using full-duplex communication. Their model, TML-Interaction-Small, claims a 0.40-second response time, comparable to natural human conversation and reportedly faster than similar models from OpenAI and Google. The approach aims to make interactivity native to the model rather than an add-on. A limited research preview is expected in the coming months, with a wider release later this year.
2 Impressions