Lance: Unified Multimodal Modeling by Multi-Task Synergy

2026-05-18

AI Summary

Existing open-source unified multimodal models struggle to simultaneously support image and video understanding, generation, and editing. This work proposes a lightweight, natively unified multimodal model grounded in two core design principles: unified context modeling and decoupled capability pathways. The architecture employs a dual-stream mixture-of-experts framework augmented with modality-aware rotary positional encoding. Through a staged multi-task training strategy combined with an adaptive data scheduling mechanism, the model achieves synergistic optimization across diverse tasks. Experimental results demonstrate that the proposed approach significantly outperforms current open-source unified models in both image and video generation while maintaining strong multimodal understanding capabilities.

Abstract

We present Lance, a lightweight native unified model supporting multimodal understanding, generation, and editing for both images and videos. Rather than relying on model capacity scaling or text-image-dominant designs, Lance explores a practical paradigm for unified multimodal modeling via collaborative multi-task training. It is grounded in two core principles: unified context modeling and decoupled capability pathways. Specifically, Lance is trained from scratch and employs a dual-stream mixture-of-experts architecture on shared interleaved multimodal sequences, enabling joint context learning while decoupling the pathways for understanding and generation. We further introduce modality-aware rotary positional encoding to mitigate interference among heterogeneous visual tokens and boost cross-task alignment. During training, Lance adopts a staged multi-task training paradigm with capability-oriented objectives and adaptive data scheduling to strengthen both semantic comprehension and visual generation performance. Experimental results demonstrate that Lance substantially outperforms existing open-source unified models in image and video generation, while retaining strong multimodal understanding capabilities. The homepage is available at https://lance-project.github.io.

Problem

• unified multimodal modeling
• multimodal understanding
• image and video generation
• multi-task synergy
• visual content editing

Innovation

• unified multimodal modeling
• multi-task synergy
• decoupled capability pathways
• modality-aware rotary positional encoding
• mixture-of-experts architecture