Consistency Should Be the Priority for Unified Multimodal Models

Feb 3, 2026·
Zhaolong Su
,
Yinyi Luo
Yiqiao Jin
Yiqiao Jin
,
Mengqi Zhang
,
Wenyue Hua
,
Srijan Kumar
,
Qingsong Wen
,
Jindong Wang
· 1 min read
Abstract
This position paper makes consistency a central objective for unified multimodal models. It outlines five connected research directions: grounding that accounts for differences in modality reliability, representations that coordinate understanding and generation, consistent multi-step reasoning, post-training for competing objectives, and evaluation of consistency across tasks.
Type
Publication
TechRxiv preprint

Overview

This position paper makes consistency a central objective for unified multimodal models. It outlines five connected research directions: grounding that accounts for differences in modality reliability, representations that coordinate understanding and generation, consistent multi-step reasoning, post-training for competing objectives, and evaluation of consistency across tasks.

Multimodal Models Position Paper Consistency Evaluation
Yiqiao Jin
Authors
Yiqiao Jin
Final-year Ph.D. Candidate in Computer Science
My research sits at the intersection of multimodal foundation models, intelligent agents, and spatial/physical intelligence. I build general-purpose models that perceive, reason, learn, and act in interactive virtual and physical environments.