---
title: "FAER — paper by Song et al. (2026) — summary and ideas"
description: "Embodied agents increasingly rely on vision-language models (VLMs) to perceive their surroundings and choose actions, so… 27 references."
authors: ["Hao Song", "Kaifeng Liu", "Yuanxing Liu", "Xiang Tian", "Xuesong Wang", "Chen Yifan", "Weinan Zhang", "Ting Liu"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.948/"
doi: "10.18653/v1/2026.findings-acl.948"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/song-2026-faer-benchmarking-vlms-failure-aware/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/song-2026-faer-benchmarking-vlms-failure-aware/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/29/) / FAER: Benchmarking VLMs for Failure-Aware Embodied Reasoning

# FAER: Benchmarking VLMs for Failure-Aware Embodied Reasoning

**Authors:** Hao Song, Kaifeng Liu, Yuanxing Liu, Xiang Tian, Xuesong Wang, Chen Yifan, [Weinan Zhang](https://md.research.elamur.ai/free/en/authors/weinan-zhang/) and [Ting Liu](https://md.research.elamur.ai/free/en/authors/ting-liu/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.948/) · [PDF](https://aclanthology.org/2026.findings-acl.948.pdf) · [DOI 10.18653/v1/2026.findings-acl.948](https://doi.org/10.18653/v1/2026.findings-acl.948) · [OpenAlex](https://openalex.org/W7166842421)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Embodied agents increasingly rely on vision-language models (VLMs) to perceive their surroundings and choose actions, so they must notice when something goes wrong. Earlier benchmarks such as AHA, RoboFAC and RoboFail mostly test fine-grained manipulation errors on a fixed tabletop. They largely ignore higher-level mishaps in long, multi-room household tasks. FAER fills this gap. The authors built FAER-VirtualHome, a pipeline that takes VirtualHome scripts, plus extra scripts generated by an LLM, and injects failures with rule-based logic. The failures are dropped objects, failed searches and failed grasps, each paired with a recovery action. The pipeline then renders synchronized first- and third-person videos together with per-action state snapshots. Each episode supports four chained tasks: detection, categorization, description and correction. Across 16 VLMs and 4 LLMs, detection proved the weakest link: models, including GPT-4o, frequently overlook abnormal events. LLMs that receive privileged simulator state as text do better. The benchmark gives a concrete tool for improving how robustly agents perceive failures.

## Key points
- The dataset contains 3,323 episodes covering 3 scenes, 65 tasks and 83 objects.
  - By failure type, there are about 1.17k drop, 1.51k find-failure and 0.62k grab-failure trajectories.
  - It also includes random normal samples.
  - There are roughly 12k Q&A pairs, about 8k for training and 3k for testing.
- The test split is balanced between failure and normal cases and includes unseen scenes.
- The best configuration was GPT-4o with 4 images, at 85.43% average accuracy.
  - Removing images lowered this to 75.69%.
  - Adding text-described states recovered part of that loss, especially for detection.
- Removing earlier-stage answers from the prompt (independent mode) reduced GPT-4o's average by 16.54%. Cascaded reasoning therefore helps the downstream correction task.
- GPT-4o's errors are concentrated in detection. In categorization, it tends to mislabel failures as "find_failed."

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works · [⚙️ Engineering](https://md.research.elamur.ai/free/en/shelves/engineering/) · 133 works · [⚙️ Control and Systems Engineering](https://md.research.elamur.ai/free/en/shelves/engineering/control-and-systems-engineering/) · 45 works

🧭 Topics (1): [🧮 Reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/) · 1,426 works

## Cite

Cite as: Hao Song, Kaifeng Liu, Yuanxing Liu, Xiang Tian, Xuesong Wang, Chen Yifan, Weinan Zhang and Ting Liu. 2026. FAER: Benchmarking VLMs for Failure-Aware Embodied Reasoning. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{song2026faer,
  title = {{FAER: Benchmarking VLMs for Failure-Aware Embodied Reasoning}},
  author = {Hao Song, Kaifeng Liu, Yuanxing Liu, Xiang Tian, Xuesong Wang, Chen Yifan, Weinan Zhang and Ting Liu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.948},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.948}
}
```

- [← previous FADE: Probing the Limits of VLMs on fine-grained OCR](https://md.research.elamur.ai/free/en/papers/shah-2026-fade-probing-limits-vlms-fine/)
- [next → FAIRGAMER: Evaluating Social Biases in LLM-Based Video Game NPCs](https://md.research.elamur.ai/free/en/papers/shi-2026-fairgamer-evaluating-social-biases-llm/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/29/)

## References

27 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. AI@Meta. 2024. Llama 3 model card. [find](https://scholar.google.com/scholar?q=AI%40Meta%2E+2024%2E+Llama+3+model+card%2E)
2. Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhaohai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, and 8 others. 2025. [Qwen2.5-vl technical report](https://md.research.elamur.ai/free/en/papers/bai-2025-qwen2-5-vl-technical/). abs/2502.13923.
3. Sijie Cheng, Zhicheng Guo, Jingwen Wu, Kechen Fang, Peng Li, Huaping Liu, and Yang Liu. 2024. Egothink: Evaluating first-person perspective thinking capability of vision-language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14291–14302. [find](https://scholar.google.com/scholar?q=Sijie+Cheng%2C+Zhicheng+Guo%2C+Jingwen+Wu%2C+Kechen+Fang%2C+Peng+Li%2C+Huaping+Liu%2C+and+Yang+Liu%2E+2024%2E+Egothink%3A+Evaluating+first%2Dperson+perspective+thinking+capability+of+vision%2Dlanguage+models%2E+In+Proceeding)
4. Yinpei Dai, Jayjun Lee, Nima Fazeli, and Joyce Chai. 2024. Racer: Rich language-guided failure recovery policies for imitation learning. arXiv preprint arXiv:2409.14674. — [arXiv:2409.14674](https://arxiv.org/abs/2409.14674) [find](https://scholar.google.com/scholar?q=Yinpei+Dai%2C+Jayjun+Lee%2C+Nima+Fazeli%2C+and+Joyce+Chai%2E+2024%2E+Racer%3A+Rich+language%2Dguided+failure+recovery+policies+for+imitation+learning%2E+arXiv+preprint+arXiv%3A2409%2E14674%2E+%E2%80%94+%5BarXiv%3A2409%2E14674%5D%28https%3A%2F%2Fa)
5. Yan Ding, Xiaohan Zhang, Saeid Amiri, Nieqing Cao, Hao Yang, Andy Kaminski, Chad Esselink, and Shiqi Zhang. 2023. Integrating action knowledge and llms for task planning and situation handling in open worlds. Auton. Robots, 47(8):981–997. [find](https://scholar.google.com/scholar?q=Yan+Ding%2C+Xiaohan+Zhang%2C+Saeid+Amiri%2C+Nieqing+Cao%2C+Hao+Yang%2C+Andy+Kaminski%2C+Chad+Esselink%2C+and+Shiqi+Zhang%2E+2023%2E+Integrating+action+knowledge+and+llms+for+task+planning+and+situation+handling+in+open)
6. Danny Driess, Fei Xia, Mehdi SM Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, and 1 others. 2023. Palm-e: an embodied multimodal language model. In Proceedings of the 40th International Conference on Machine Learning, pages 8469–8488. [find](https://scholar.google.com/scholar?q=Danny+Driess%2C+Fei+Xia%2C+Mehdi+SM+Sajjadi%2C+Corey+Lynch%2C+Aakanksha+Chowdhery%2C+Brian+Ichter%2C+Ayzaan+Wahid%2C+Jonathan+Tompson%2C+Quan+Vuong%2C+Tianhe+Yu%2C+and+1+others%2E+2023%2E+Palm%2De%3A+an+embodied+multimodal+langu)
7. Jiafei Duan, Wilbert Pumacay, Nishanth Kumar, Yi Ru Wang, Shulin Tian, Wentao Yuan, Ranjay Krishna, Dieter Fox, Ajay Mandlekar, and Yijie Guo. 2025. AHA: A vision-language-model for detecting and reasoning over failures in robotic manipulation. In The Thirteenth International Conference on Learning… [find](https://scholar.google.com/scholar?q=Jiafei+Duan%2C+Wilbert+Pumacay%2C+Nishanth+Kumar%2C+Yi+Ru+Wang%2C+Shulin+Tian%2C+Wentao+Yuan%2C+Ranjay+Krishna%2C+Dieter+Fox%2C+Ajay+Mandlekar%2C+and+Yijie+Guo%2E+2025%2E+AHA%3A+A+vision%2Dlanguage%2Dmodel+for+detecting+and+reas)
8. Zane Durante, Qiuyuan Huang, Naoki Wake, Ran Gong, Jae Sung Park, Bidipta Sarkar, Rohan Taori, Yusuke Noda, Demetri Terzopoulos, Yejin Choi, Katsushi Ikeuchi, Hoi Vo, Li Fei-Fei, and Jianfeng Gao. 2024. Agent ai: Surveying the horizons of multimodal interaction. abs/2401.03568. — [arXiv:2401.03568](https://arxiv.org/abs/2401.03568) [find](https://scholar.google.com/scholar?q=Zane+Durante%2C+Qiuyuan+Huang%2C+Naoki+Wake%2C+Ran+Gong%2C+Jae+Sung+Park%2C+Bidipta+Sarkar%2C+Rohan+Taori%2C+Yusuke+Noda%2C+Demetri+Terzopoulos%2C+Yejin+Choi%2C+Katsushi+Ikeuchi%2C+Hoi+Vo%2C+Li+Fei%2DFei%2C+and+Jianfeng+Gao%2E+202)
9. Yanjiang Guo, Yen-Jen Wang, Lihan Zha, and Jianyu Chen. 2024. Doremi: Grounding language model by detecting and recovering from plan-execution misalignment. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 12124–12131. IEEE. [find](https://scholar.google.com/scholar?q=Yanjiang+Guo%2C+Yen%2DJen+Wang%2C+Lihan+Zha%2C+and+Jianyu+Chen%2E+2024%2E+Doremi%3A+Grounding+language+model+by+detecting+and+recovering+from+plan%2Dexecution+misalignment%2E+In+2024+IEEE%2FRSJ+International+Conference+o)
10. Baoxiong Jia, Ting Lei, Song-Chun Zhu, and Siyuan Huang. 2022. Egotaskqa: Understanding human tasks in egocentric videos. In Advances in Neural Information Processing Systems, volume 35, pages 3343–3360. Curran Associates, Inc. [find](https://scholar.google.com/scholar?q=Baoxiong+Jia%2C+Ting+Lei%2C+Song%2DChun+Zhu%2C+and+Siyuan+Huang%2E+2022%2E+Egotaskqa%3A+Understanding+human+tasks+in+egocentric+videos%2E+In+Advances+in+Neural+Information+Processing+Systems%2C+volume+35%2C+pages+3343%E2%80%9333)
11. Eric Kolve, Roozbeh Mottaghi, Winson Han, Eli VanderBilt, Luca Weihs, Alvaro Herrasti, Matt Deitke, Kiana Ehsani, Daniel Gordon, Yuke Zhu, Aniruddha Kembhavi, Abhinav Gupta, and Ali Farhadi. 2017. Ai2-thor: an interactive 3d environment for visual ai. abs/1712.05474. [find](https://scholar.google.com/scholar?q=Eric+Kolve%2C+Roozbeh+Mottaghi%2C+Winson+Han%2C+Eli+VanderBilt%2C+Luca+Weihs%2C+Alvaro+Herrasti%2C+Matt+Deitke%2C+Kiana+Ehsani%2C+Daniel+Gordon%2C+Yuke+Zhu%2C+Aniruddha+Kembhavi%2C+Abhinav+Gupta%2C+and+Ali+Farhadi%2E+2017%2E+Ai2)
12. Mingcong Lei, Ge Wang, Yiming Zhao, Zhixin Mai, Qing Zhao, Yao Guo, Zhen Li, Shuguang Cui, Yatong Han, and Jinke Ren. 2025. Clea: Closed-loop embodied agent for enhancing task execution in dynamic environments. abs/2503.00729. — [arXiv:2503.00729](https://arxiv.org/abs/2503.00729) [find](https://scholar.google.com/scholar?q=Mingcong+Lei%2C+Ge+Wang%2C+Yiming+Zhao%2C+Zhixin+Mai%2C+Qing+Zhao%2C+Yao+Guo%2C+Zhen+Li%2C+Shuguang+Cui%2C+Yatong+Han%2C+and+Jinke+Ren%2E+2025%2E+Clea%3A+Closed%2Dloop+embodied+agent+for+enhancing+task+execution+in+dynamic+env)
13. Boyi Li, Philipp Wu, Pieter Abbeel, and Jitendra Malik. 2025a. Interactive task planning with language models. Transactions on Machine Learning Research. [find](https://scholar.google.com/scholar?q=Boyi+Li%2C+Philipp+Wu%2C+Pieter+Abbeel%2C+and+Jitendra+Malik%2E+2025a%2E+Interactive+task+planning+with+language+models%2E+Transactions+on+Machine+Learning+Research%2E)
14. Chengshu Li, Ruohan Zhang, Josiah Wong, Cem Gokmen, Sanjana Srivastava, Roberto Martín-Martín, Chen Wang, Gabrael Levine, Michael Lingelbach, Jiankai Sun, Mona Anvari, Minjune Hwang, Manasi Sharma, Arman Aydin, Dhruva Bansal, Samuel Hunter, Kyu-Young Kim, Alan Lou, Caleb R Matthews, and 9 others. 20… — [arXiv:2503.08604](https://arxiv.org/abs/2503.08604) [find](https://scholar.google.com/scholar?q=Chengshu+Li%2C+Ruohan+Zhang%2C+Josiah+Wong%2C+Cem+Gokmen%2C+Sanjana+Srivastava%2C+Roberto+Mart%C3%ADn%2DMart%C3%ADn%2C+Chen+Wang%2C+Gabrael+Levine%2C+Michael+Lingelbach%2C+Jiankai+Sun%2C+Mona+Anvari%2C+Minjune+Hwang%2C+Manasi+Sharma%2C+Ar)
15. Manling Li, Shiyu Zhao, Qineng Wang, Kangrui Wang, Yu Zhou, Sanjana Srivastava, Cem Gokmen, Tony Lee, Li Erran Li, Ruohan Zhang, Weiyu Liu, Percy Liang, Li Fei-Fei, Jiayuan Mao, and Jiajun Wu. 2024b. Embodied agent interface: Benchmarking LLMs for embodied decision making. In The Thirty-eight Confer… [find](https://scholar.google.com/scholar?q=Manling+Li%2C+Shiyu+Zhao%2C+Qineng+Wang%2C+Kangrui+Wang%2C+Yu+Zhou%2C+Sanjana+Srivastava%2C+Cem+Gokmen%2C+Tony+Lee%2C+Li+Erran+Li%2C+Ruohan+Zhang%2C+Weiyu+Liu%2C+Percy+Liang%2C+Li+Fei%2DFei%2C+Jiayuan+Mao%2C+and+Jiajun+Wu%2E+2024b%2E+)
16. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024. [Improved baselines with visual instruction tuning](https://md.research.elamur.ai/free/en/papers/liu-2023-improved-baselines-visual-instruction-tuning/). In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 26296–26306.
17. Zeyi Liu, Arpit Bahety, and Shuran Song. 2023. Reflect: Summarizing robot experiences for failure explanation and correction. In Proceedings of The 7th Conference on Robot Learning, volume 229 of Proceedings of Machine Learning Research, pages 3468– 3484. PMLR. [find](https://scholar.google.com/scholar?q=Zeyi+Liu%2C+Arpit+Bahety%2C+and+Shuran+Song%2E+2023%2E+Reflect%3A+Summarizing+robot+experiences+for+failure+explanation+and+correction%2E+In+Proceedings+of+The+7th+Conference+on+Robot+Learning%2C+volume+229+of+Proc)
18. Weifeng Lu, Minghao Ye, Zewei Ye, Ruihan Tao, Shuo Yang, and Bo Zhao. 2025. Robofac: A comprehensive framework for robotic failure analysis and correction. Preprint, arXiv:2505.12224. — [arXiv:2505.12224](https://arxiv.org/abs/2505.12224) [find](https://scholar.google.com/scholar?q=Weifeng+Lu%2C+Minghao+Ye%2C+Zewei+Ye%2C+Ruihan+Tao%2C+Shuo+Yang%2C+and+Bo+Zhao%2E+2025%2E+Robofac%3A+A+comprehensive+framework+for+robotic+failure+analysis+and+correction%2E+Preprint%2C+arXiv%3A2505%2E12224%2E+%E2%80%94+%5BarXiv%3A2505%2E12)
19. Xavier Puig, Kevin Ra, Marko Boben, Jiaman Li, Tingwu Wang, Sanja Fidler, and Antonio Torralba. 2018. Virtualhome: Simulating household activities via programs. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). [find](https://scholar.google.com/scholar?q=Xavier+Puig%2C+Kevin+Ra%2C+Marko+Boben%2C+Jiaman+Li%2C+Tingwu+Wang%2C+Sanja+Fidler%2C+and+Antonio+Torralba%2E+2018%2E+Virtualhome%3A+Simulating+household+activities+via+programs%2E+In+Proceedings+of+the+IEEE+Conference+o)
20. Mohit Shridhar, Jesse Thomason, Daniel Gordon, Yonatan Bisk, Winson Han, Roozbeh Mottaghi, Luke Zettlemoyer, and Dieter Fox. 2020. [ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Tasks](https://md.research.elamur.ai/free/en/papers/shridhar-2019-alfred-benchmark-interpreting-grounded-instructions/). In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
21. Qwen Team. 2024. Qwen2.5: A party of foundation models. [find](https://scholar.google.com/scholar?q=Qwen+Team%2E+2024%2E+Qwen2%2E5%3A+A+party+of+foundation+models%2E)
22. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Yang Fan, Kai Dang, Mengfei Du, Xuancheng Ren, Rui Men, Dayiheng Liu, Chang Zhou, Jingren Zhou, and Junyang Lin. 2024a. [Qwen2-vl: Enhancing vision-language model’s perception of the…](https://md.research.elamur.ai/free/en/papers/wang-2024-qwen2-vl-enhancing-vision/)
23. Chenxuan Li, Jiaming Liu, Guanqun Wang, Xiaoqi Li, Sixiang Chen, Liang Heng, Chuyan Xiong, Jiaxin Ge, Renrui Zhang, Kaichen Zhou, and Shanghang Zhang. 2024a. A self-correcting vision-language-action model for fast and slow system manipulation. abs/2405.17418. — [arXiv:2405.17418](https://arxiv.org/abs/2405.17418) [find](https://scholar.google.com/scholar?q=Chenxuan+Li%2C+Jiaming+Liu%2C+Guanqun+Wang%2C+Xiaoqi+Li%2C+Sixiang+Chen%2C+Liang+Heng%2C+Chuyan+Xiong%2C+Jiaxin+Ge%2C+Renrui+Zhang%2C+Kaichen+Zhou%2C+and+Shanghang+Zhang%2E+2024a%2E+A+self%2Dcorrecting+vision%2Dlanguage%2Daction+m)
24. Weiyun Wang, Zhe Chen, Wenhai Wang, Yue Cao, Yangzhou Liu, Zhangwei Gao, Jinguo Zhu, Xizhou Zhu, Lewei Lu, Yu Qiao, and Jifeng Dai. 2024b. [Enhancing the reasoning ability of multimodal large language models via mixed preference optimization](https://md.research.elamur.ai/free/en/papers/wang-2024-enhancing-reasoning-ability-multimodal-large/). abs/2411.10442.
25. Dongping Li, Tielong Cai, Tianci Tang, Wenhao Chai, Katherine Rose Driggs-Campbell, and Gaoang Wang. 2025b. Emmoe: A comprehensive benchmark for Rui Yang, Hanyang Chen, Junyu Zhang, Mark Zhao, Cheng Qian, Kangrui Wang, Qineng Wang, Teja Venkat Koripella, Marziyeh Movahedi, Manling Li, and 1 others.… [find](https://scholar.google.com/scholar?q=Dongping+Li%2C+Tielong+Cai%2C+Tianci+Tang%2C+Wenhao+Chai%2C+Katherine+Rose+Driggs%2DCampbell%2C+and+Gaoang+Wang%2E+2025b%2E+Emmoe%3A+A+comprehensive+benchmark+for+Rui+Yang%2C+Hanyang+Chen%2C+Junyu+Zhang%2C+Mark+Zhao%2C+Cheng+Q)
26. Jianguo Zhang, Tian Lan, Ming Zhu, Zuxin Liu, Thai Hoang, Shirley Kokane, Weiran Yao, Juntao Tan, Akshara Prabhakar, Haolin Chen, Zhiwei Liu, Yihao Feng, Tulika Awalgaonkar, Rithesh Murthy, Eric Hu, Zeyuan Chen, Ran Xu, Juan Carlos Niebles, Shelby Heinecke, and 3 others. 2025. [Xlam: A family of larg…](https://md.research.elamur.ai/free/en/papers/zhang-2025-xlam-family-large-action-models/)
27. Jinguo Zhu, Weiyun Wang, Zhe Chen, Zhaoyang Liu, Shenglong Ye, Lixin Gu, Hao Tian, Yuchen Duan, Weijie Su, Jie Shao, Zhangwei Gao, Erfei Cui, Xuehui Wang, Yue Cao, Yangzhou Liu, Xingguang Wei, Hongjie Zhang, Haomin Wang, Weiye Xu, and 32 others. 2025. [Internvl3: Exploring advanced training and test-…](https://md.research.elamur.ai/free/en/papers/zhu-2025-internvl3-exploring-advanced-training/)

*Updated 1 November 2026*