Unverified50% confidenceReleaseExact time
VIVX has released the world's first real-time interactive multimodal models A1 and W1, with approximately 30 billion parameters, capable of interactive use while generating video.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/12/2026
First Seen
Valid until: 12/11/2026
Sources
Related Entities
Related Claims
Unverified昆仑万维开源升级了世界模型Matrix-3.5，支持实时生成可自由探索的虚拟场景和多人称视角67% similarUnverifiedVideo VQ-VAE, 3D convolution, and encoding uniformly sampled keyframes with ViT are the mainstream video tokenization methods.64% similarUnverifiedMOSS-VL-Realtime adopts an online processing paradigm, capable of continuously understanding frames and answering questions in real time during video playback, unlike the offline processing paradigm of traditional VideoQA models63% similarUnverifiedThe open-source community released a VR180 stereoscopic video LoRA model customized for the MiniMax H3 video generation model, capable of generating VR180 stereoscopic content in side-by-side (SBS) format.63% similarUnverified原生视频理解模型主流架构包括基于3D卷积的时空特征提取（如C3D、I3D）、基于视频Transformer的注意力建模（如Video Swin Transformer、TimeSformer）和视频版视觉语言模型（如Gemini 1.5、MovieChat）62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/911523API
curl https://kongchang.com/api/v1/knowledge/claims/911523MCP
get_claim(id=911523)