ARTFEED — Contemporary Art Intelligence

Thinking Machines Unveils AI That Can Interrupt You in Conversation

ai-technology · 2026-05-12

On Monday, Thinking Machines Lab, an AI startup established last year by former OpenAI CTO Mira Murati, unveiled a new AI model named TML-Interaction-Small. This model features the ability to process user inputs and generate responses at the same time, facilitating a more natural interaction akin to a phone call. This capability, referred to as “full duplex,” allows for interruptions, unlike existing models that function in a turn-taking manner. The company boasts a response time of 0.40 seconds, comparable to natural human dialogue and significantly quicker than similar offerings from OpenAI and Google. Currently, this release is a research preview, with a limited version anticipated in the coming months and a broader rollout later this year.

Key facts

  • Thinking Machines Lab was founded last year by former OpenAI CTO Mira Murati.
  • The company announced interaction models that enable full-duplex communication.
  • The model is named TML-Interaction-Small.
  • Response time is 0.40 seconds, faster than OpenAI and Google models.
  • The release is a limited research preview coming in the next few months.
  • A wider release is planned for later this year.
  • Full duplex means the AI can process input and generate output simultaneously.
  • Current AI models operate in a turn-based manner.

Entities

Institutions

  • Thinking Machines Lab
  • OpenAI
  • Google

Sources