VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs
6 Oct 2026
Multimodal large language models have become the dominant paradigm for visual understanding, but incur substantial costs by encoding inputs into dense, fixed-size patch tokens.