The Sequence AI of the Week #863: The Model is the Interface: Inside Thinking Machines' Interactive Models

2026-05-20 · Source: TheSequence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

Thinking Machines is developing interactive models that redefine the interaction paradigm for large language models, moving beyond the conventional "concatenate tokens, predict next token" approach. This work aims to elevate multi-modality, recognizing that collaborative interactions are inherently temporal, unlike the buffered and serialized nature of text-based communication. The current default mental model for LLMs, where humans write a message and the model replies, is effective for many text-centric tasks due to text's forgiving nature. However, Thinking Machines' early efforts suggest a shift towards a more dynamic, real-time engagement where the model itself functions as the interface, facilitating a more natural and temporal collaborative experience.

Key takeaway

For AI Architects designing next-generation human-AI collaboration systems, recognize that traditional text-based LLM interfaces may limit true interactive potential. You should explore "model as interface" concepts, like those from Thinking Machines, to enable more temporal and multi-modal interactions. This shift could inform your strategic planning for applications requiring real-time, dynamic collaboration, moving beyond simple turn-taking conversational models.

Key insights

Thinking Machines proposes "the model is the interface" for temporal, collaborative AI interactions, transcending sequential text processing.

Principles

• The model serves as the interface.
• Collaboration demands temporal interaction.
• Text-based LLMs are sequentially constrained.

Topics

• Thinking Machines
• Interactive AI Models
• Multi-modality
• LLM Interfaces
• Human-AI Collaboration
• Temporal AI

Best for: Research Scientist, AI Product Manager, AI Scientist, AI Architect, Director of AI/ML

Related on AIssential

• Interaction Models: A Scalable Approach to Human-AI Collaboration — Native, real-time multimodal interaction within AI models enhances human-AI collaboration beyond turn-based systems.
• New AI Model SHOCKED Me — Thinking Machines Labs' new model offers real-time, multimodal, and context-aware conversational AI.
• [AINews] Thinking Machines' Native Interaction Models - TML-Interaction-Small 276B-A12B - advances SOTA Realtime Voice and kills standard VAD — Real-time, full-duplex multimodal interaction models are advancing human-AI collaboration beyond turn-based systems.
• Nobody gets this right — The distinction between language models and world models is diminishing as AI becomes multimodal and adept at abstract mathematical representations.
• Inside Thinking Machines’ Interaction Models — Integrating interactivity directly into AI models, rather than relying on external "harness" components, unlocks true real-time human-AI collaboration.
• Mira Murati steps back into the spotlight, carefully — Thinking Machines Lab is developing "interaction models" for real-time, continuous human-AI communication beyond turn-based prompts.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TheSequence.