AEGIS: Exploring the Limit of World Knowledge Capabilities for Unified Mulitmodal Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909979912437760 |
|---|---|
| author | Lin, Jintao Dong, Bowen Shi, Weikang Lei, Chenyang Zhang, Suiyun Liu, Rui Liu, Xihui |
| author_facet | Lin, Jintao Dong, Bowen Shi, Weikang Lei, Chenyang Zhang, Suiyun Liu, Rui Liu, Xihui |
| contents | The capability of Unified Multimodal Models (UMMs) to apply world knowledge across diverse tasks remains a critical, unresolved challenge. Existing benchmarks fall short, offering only siloed, single-task evaluations with limited diagnostic power. To bridge this gap, we propose AEGIS (\emph{i.e.}, \textbf{A}ssessing \textbf{E}diting, \textbf{G}eneration, \textbf{I}nterpretation-Understanding for \textbf{S}uper-intelligence), a comprehensive multi-task benchmark covering visual understanding, generation, editing, and interleaved generation. AEGIS comprises 1,050 challenging, manually-annotated questions spanning 21 topics (including STEM, humanities, daily life, etc.) and 6 reasoning types. To concretely evaluate the performance of UMMs in world knowledge scope without ambiguous metrics, we further propose Deterministic Checklist-based Evaluation (DCE), a protocol that replaces ambiguous prompt-based scoring with atomic ``Y/N'' judgments, to enhance evaluation reliability. Our extensive experiments reveal that most UMMs exhibit severe world knowledge deficits and that performance degrades significantly with complex reasoning. Additionally, simple plug-in reasoning modules can partially mitigate these vulnerabilities, highlighting a promising direction for future research. These results highlight the importance of world-knowledge-based reasoning as a critical frontier for UMMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_00561 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | AEGIS: Exploring the Limit of World Knowledge Capabilities for Unified Mulitmodal Models Lin, Jintao Dong, Bowen Shi, Weikang Lei, Chenyang Zhang, Suiyun Liu, Rui Liu, Xihui Computer Vision and Pattern Recognition The capability of Unified Multimodal Models (UMMs) to apply world knowledge across diverse tasks remains a critical, unresolved challenge. Existing benchmarks fall short, offering only siloed, single-task evaluations with limited diagnostic power. To bridge this gap, we propose AEGIS (\emph{i.e.}, \textbf{A}ssessing \textbf{E}diting, \textbf{G}eneration, \textbf{I}nterpretation-Understanding for \textbf{S}uper-intelligence), a comprehensive multi-task benchmark covering visual understanding, generation, editing, and interleaved generation. AEGIS comprises 1,050 challenging, manually-annotated questions spanning 21 topics (including STEM, humanities, daily life, etc.) and 6 reasoning types. To concretely evaluate the performance of UMMs in world knowledge scope without ambiguous metrics, we further propose Deterministic Checklist-based Evaluation (DCE), a protocol that replaces ambiguous prompt-based scoring with atomic ``Y/N'' judgments, to enhance evaluation reliability. Our extensive experiments reveal that most UMMs exhibit severe world knowledge deficits and that performance degrades significantly with complex reasoning. Additionally, simple plug-in reasoning modules can partially mitigate these vulnerabilities, highlighting a promising direction for future research. These results highlight the importance of world-knowledge-based reasoning as a critical frontier for UMMs. |
| title | AEGIS: Exploring the Limit of World Knowledge Capabilities for Unified Mulitmodal Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2601.00561 |