Mira Murati Spotlights Thinking Machines' Streaming Multimodal AI Interface
Former OpenAI CTO Mira Murati previewed Thinking Machines' new streaming multimodal interface, which processes audio, text, and video in 200-millisecond intervals to enable continuous human-machine collaboration. The system is still in early development, with no firm release dates, and represents Murati's first major media appearance in 18 months, during which she also addressed prior internal OpenAI turmoil.
Key Takeaways
- Thinking Machines' interface processes continuous streams of audio, text, and video in 200-millisecond intervals.
- The system focuses on real-time conversational dynamics, including interruptions and mid-thought corrections.
- Murati declined to provide firm release dates, characterizing the effort as an early-stage product.
- This marks Murati's first major media appearance in 18 months, addressing prior OpenAI internal turmoil.
- Murati advocates for advanced AI development through sustained human-machine collaboration, not occasional oversight.
Why It Matters
This initiative signals a focus on low-latency, real-time multimodal AI, moving beyond turn-based interactions currently prevalent in many systems. For the streaming ecosystem, improved human-machine collaboration in AI could translate to more dynamic content creation tools, enhanced viewer engagement analytics, and more natural interactive experiences. What to watch: Thinking Machines' technical publications and API documentation for evaluation metrics on robustness to interruptions and context retention in streaming scenarios.
Additional Context
Mira Murati, in her Bloomberg interview on June 4, 2026, as reported by TechCrunch and WebProNews on June 5, stepped back into public view after 18 months to discuss Thinking Machines Lab's latest developments. The company's "interaction models" aim for "full duplex" communication, allowing AI to process input and generate responses simultaneously, akin to natural human conversation (TechCrunch, May 2026). Thinking Machines claims its TML-Interaction-Small model responds in 0.40 seconds, matching the speed of natural human conversation and significantly faster than current models from OpenAI and Google (WebProNews, June 2026). While described as a research preview, not a finished product, a limited research preview is expected in the coming months, with broader access aimed for later in 2026 (The Economic Times, June 2026). Murati also discussed the November 2023 OpenAI board changes, referring to it internally as "the blip," and emphasized the need for improved checks and balances in AI governance (TechCrunch, June 2026).
Read full article at letsdatascience.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source