Thinking Machines' preview real-time model that processes audio, video and text continuously while responding.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Interaction Models are real-time multimodal models from Thinking Machines Lab, the AI company founded by Mira Murati. According to creator coverage of the announcement, they take in audio, video and text continuously while thinking and responding, rather than working in strict turns. This lets them handle interruptions and visual cues, and speak when appropriate.
Demonstrations reportedly included simultaneous translation while a person is still speaking, counting specific words, tracking time, and making tool calls in parallel with the conversation. A lightweight interaction model manages the live exchange and hands off to a background model for deeper reasoning.
The models are not publicly available. A limited research preview was said to be planned for the coming months. No pricing, API, documentation, weights or paper were found in the provided material, so details such as supported languages and limits are unconfirmed.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Real-time models that take in audio, video and text continuously while thinking and responding, handling interruptions and visual cues. A lightweight interaction model hands off to a background model for deeper reasoning. A limited research preview is planned in the coming months.
Preview model from Mira Murati's lab that handles real-time conversation: translates while the speaker talks, interrupts when appropriate, counts animal words, tracks time, and does simultaneous tool calls. Not publicly available; limited research preview planned in coming months.
Closest neighbours by meaning, not just by category.
Model · Block
Block research preview of a full-duplex speech-to-speech model trained only on human conversation audio, without text.
Model
Research model for real-time, talking video agents with a single streaming Transformer; 640×368 output at ~200 ms latency.
Model · Alibaba
Alibaba's real-time audio-visual interactive character video model that renders speech and free-form actions.
Model · Google
Google's real-time voice model with an Extended Thinking variant, available in the Gemini app, AI Studio and Search Live.
Model · Thinking Machines
Open-weights multimodal MoE model (276B total, 12B active) from Thinking Machines; text, image and audio in, text out.
Model · Google
Google's low-latency, bidirectional voice and audio streaming models, accessed through the Gemini Live API.
