JoyAI-VL-Interaction: Real-Time Vision-Language Interaction Intelligence

2026-06-09

AI Summary

This work addresses the limitation of existing large models, which are typically confined to turn-based interactions and struggle to determine in real time whether to intervene in dynamic visual environments. The authors propose the first open-source, vision-driven real-time interactive system, built upon an 8B-parameter vision-first multimodal architecture that continuously processes video streams and autonomously decides to remain silent, respond directly, or invoke backend agents. The system integrates automatic speech recognition (ASR), text-to-speech (TTS), a memory module, a visual user interface, and a plug-and-play agent interface, enabling emergent capabilities—such as cross-screen shopping assistance and impromptu scene narration—that were not explicitly trained. Evaluated across six real-world scenarios, the system significantly outperforms Doubao and Gemini video assistants and is strongly preferred by human evaluators.

Abstract

Many moments in the real world do not wait for a user to ask. A fire starts on a security monitor, an expression flickers across a video call, or a product a viewer wants flashes by in a livestream. Yet today's large models remain mostly turn-based by design: they answer only when addressed, and even video-call apps that appear interactive still operate as question-answer systems, reacting only when polled or prompted. We argue for a different paradigm: a model that is present in the world like a person. It continuously watches what is happening now, decides on its own whether to speak or stay silent, interacts in real time, and delegates to a background model when the problem is hard. To advance interaction models and their adoption across domains, we make two fully open-sourced contributions. First, we release JoyAI-VL-Interaction, an 8B-scale, vision-first VL-interaction model. The model makes the response decision internally, choosing each second to stay silent, respond, or delegate to a background model, and it excels at vision-triggered responsiveness and time awareness. We pair it with a transferable training recipe, from which capabilities we never trained for emerge, such as guiding a shopper through changing app screens or improvising a lecture from a slide deck. Second, we release a complete, deployable system built around that model. The system streams any ongoing video into the model, making it genuinely present in the world. All other components are pluggable, including ASR/TTS modules, memory, visualization UI, and a background brain that can connect to any API or agent. Across six real-world scenarios, human raters prefer JoyAI-VL-Interaction over the in-app video-call assistants of Doubao and Gemini by a wide margin. To our knowledge, this is the first open, vision-driven interaction model released together with its training recipe, data, and complete deployable system.

Problem

• real-time interaction
• vision-language models
• continuous perception
• autonomous response
• interactive AI

Innovation

• real-time interaction
• vision-language model
• autonomous response decision
• deployable AI system
• time-awareness