Mira Murati’s Thinking Machines Lab Introduces Interaction Models: A Native Multimodal Architecture for Real-Time Human-AI Collaboration
276B parameters, 12B active. Mira Murati's new model eliminates latency in real-time human-AI dialogue.
Why it matters
Thinking Machines Lab's TML-Interaction-Small demonstrates a fundamental shift in multimodal architecture—parallel real-time interaction + async reasoning—that could redefine how AI systems handle continuous, full-duplex human engagement. This is a capability benchmark story from a prominent AI founder.
The key facts
9 to know276B parameter Mixture-of-Experts model
12B active parameters
Processes 200ms chunks of audio, video, text simultaneously
Eliminates external voice-activity detection requirement
Dual-component architecture: real-time interaction model + asynchronous background model
Full-duplex exchange capability
Research preview of TML-Interaction-Small
Founder: Mira Murati (former OpenAI CTO)
Published May 13, 2026 on MarkTechPost
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Thinking Machines Lab has introduced a research preview of TML-Interaction-Small, a 276B parameter Mixture-of-Experts model with 12B active parameters, built around a multi-stream, time-aligned micro-turn architecture that processes 200ms chunks of audio, video, and text simultaneously —…