# UniG2U-Bench: Do Unified Models Advance Multimodal Understanding?

arXiv: https://arxiv.org/abs/2603.03241 · HTML: https://mlsanity.com/paper/2603.03241/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Mar 2026.

## Built on by (most cited first, 7 of 7)

- [Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and Recipes](https://mlsanity.com/paper/2608.05000/) (2026)
- [TorchUMM: A Unified Multimodal Model Codebase for Evaluation, Analysis, and Post-training](https://mlsanity.com/paper/2604.10784/) (2026)
- [ETCHR: Editing To Clarify and Harness Reasoning](https://mlsanity.com/paper/2605.23897/) (2026)
- [When Does Visual Generation Help Visual Understanding in Unified Multimodal Models?](https://mlsanity.com/paper/2608.22174/) (2026)
- [LatentUMM: Dual Latent Alignment for Unified Multimodal Models](https://mlsanity.com/paper/2605.17766/) (2026)
- [Beyond Visual CoT: Internalized Visual Thinking for Proactive Video Reasoning](https://mlsanity.com/paper/2608.15869/) (2026)
- [OmniTaskonomy: When Does Visual Generation Improve Visual Understanding?](https://mlsanity.com/paper/2609.38079/) (2026)

## Abstract (arXiv)

Unified multimodal models have recently demonstrated strong generative capabilities, yet whether and when generation improves understanding remains unclear. Existing benchmarks lack a systematic exploration of the specific tasks where generation facilitates understanding. To this end, we introduce UniG2U-Bench, a comprehensive benchmark categorizing generation-to-understanding (G2U) evaluation into 7 regimes and 30 subtasks, requiring varying degrees of implicit or explicit visual transformations. Extensive evaluation of over 30 models reveals three core findings: 1) Unified models generally underperform their base Vision-Language Models (VLMs), and Generate-then-Answer (GtA) inference typically degrades performance relative to direct inference. 2) Consistent enhancements emerge in spatial intelligence, visual illusions, or multi-round reasoning subtasks, where enhanced spatial and shape perception, as well as multi-step intermediate image states, prove beneficial. 3) Tasks with similar reasoning structures and models sharing architectures exhibit correlated behaviors, suggesting that generation-understanding coupling induces class-consistent inductive biases over tasks, pretraining data, and model architectures. These findings highlight the necessity for more diverse training data and novel paradigms to fully unlock the potential of unified multimodal modeling.