Elamur Research⌕
☰
home / papers / 2026 / ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation

ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation

Paper · 2026 · Yihua Zhang, Mingfu Liang, Jiyan Yang et al.

Summary

Current multimodal recommenders fuse images and text well, but they decide in one opaque forward pass. They cannot recognise when they are unsure and spend equal compute on easy and hard queries. ReasonRec, from Meta AI with Michigan State and UNC Chapel Hill collaborators, treats recommendation as an agent loop with three stages:

  • Observe: a pretrained LLaVA-style VLM encoder embeds item images, query text and user history.
  • Deliberate: the model is fine-tuned with Reasoning-Aware Visual Instruction Tuning (R-VIT). This recasts tasks as VQA-style prompts whose answers include chain-of-thought tokens. Training follows an evidence-horizon curriculum that gradually raises user-item sparsity.
  • Act: when its estimated confidence is low, the agent calls lightweight classical recommenders as tools.

Experiments cover five public datasets and four tasks: predicting CTR, producing explanations, and both the sequential and direct forms of recommendation. They report over 30% relative gains in HR@5 and NDCG@5 against prior multimodal systems, with gains holding in cold-start, time-shift and multi-domain settings. The work offers a blueprint for recommenders that are interpretable, aware of their own uncertainty and economical with compute.

Key points

  • Datasets: Sports, Beauty, Clothing, Toys and Pixel-1M. The sparsest, Pixel-1M, triggers the most tool calls, up to 34.1% of queries in CTR prediction. Delegation depends more on dataset sparsity than on task type.
  • Tool conflicts (Pixel-1M, sequential): the tool ensemble alone scores 58.4% and ReasonRec 69.3%. The two disagree in 13.6% of cases, and ReasonRec is right 84.2% of the time when they do. Noisy tool outputs injected during training teach it not to trust tools blindly.
  • Accuracy vs. latency: HR@5 is 0.0315 for ReasonRec, 0.0197 for VIP5 and 0.0116 for SASRec. Per-query time is 499 ms, 470 ms and 143 ms respectively. ReasonRec is therefore slightly slower than VIP5; the claimed savings come from routing up to ~35% of queries away from the VLM.
  • Ablation: removing the curriculum, the template mixture or the tools cuts performance by 10–45%. The curriculum matters most, and the tools add about 10%.

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

 Institutions

🎓University of North Carolina at Chapel HillChapel Hill · United States74 works · 23,278 citations🎓Michigan State UniversityEast Lansing · United States30 works · 1,375 citations

 Fields, topics and concepts

📚 Shelves 3The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksComputer Vision and Pattern Recognition · 3,486 worksInformation Systems · 1,730 works

🧭 Topics 2Research subjects and languages the work is about.Reasoning · 1,685 worksVision & language · 811 works

Cite

Cite as Yihua Zhang, Mingfu Liang, Jiyan Yang, Rong Jin, Wen-Yen Chen, Yiping Han, Huayu Li, Buyun Zhang, Liang Luo, Luke Simon, Sijia Liu, Tianlong Chen and Xi Liu. 2026. ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.391

@inproceedings{zhang2026reasonrec,
  title = {{ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation}},
  author = {Yihua Zhang, Mingfu Liang, Jiyan Yang, Rong Jin, Wen-Yen Chen, Yiping Han, Huayu Li, Buyun Zhang, Liang Luo, Luke Simon, Sijia Liu, Tianlong Chen and Xi Liu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.391},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.391}
}

References

68 references, 19 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katie Millican, Malcolm Reynolds, and 1 others. 2022. Flamingo: a visual language model for few-shot learning. arXiv preprint arXiv:2204.14198.
  2. Anas Awadalla, Irena Gao, Joshua Gardner, Jack Hessel, Yusuf Hanafy, Wanrong Zhu, Kalyani Marathe, Yonatan Bitton, Samir Gadre, Jenia Jitsev, Simon Kornblith, Pang Wei Koh, Gabriel Ilharco, Mitchell Wortsman, and Ludwig Schmidt. 2023. Openflamingo. find
  3. Xu Chen, Hanxiong Chen, Hongteng Xu, Yongfeng Zhang, Yixin Cao, Zheng Qin, and Hongyuan Zha. 2019. Personalized fashion recommendation with visual explanations based on multimodal attention network: Towards visually explainable recommendation. In Proceedings of the 42nd International ACM SIGIR Confe… find
  4. Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, Rohan Anil, Zakaria Haque, Lichan Hong, Vihan Jain, Xiaobing Liu, and Hemal Shah. 2016. Wide & deep learning for recommender systems. In Proceedings of the… find
  5. Weiyu Cheng, Yanyan Shen, and Linpeng Huang. 2020. Adaptive factorization network: Learning adaptive-order feature interactions. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 3609–3616. find
  6. Yu Cheng, Yunzhu Pan, Jiaqi Zhang, Yongxin Ni, Aixin Sun, and Fajie Yuan. 2023. An image dataset for benchmarking recommender systems with raw pixels. arXiv preprint arXiv:2309.06789. — [arXiv:2309.06789](https://arxiv.org/abs/2309.06789) find
  7. Xiangxiang Chu, Limeng Qiao, Xinyang Lin, Shuang Xu, Yang Yang, Yiming Hu, Fei Wei, Xinyu Zhang, Bo Zhang, Xiaolin Wei, and 1 others. 2023. Mobilevlm: A fast, reproducible and strong vision language assistant for mobile devices. arXiv preprint arXiv:2312.16886. find
  8. Paul Covington, Jay Adams, and Emre Sargin. 2016. Deep neural networks for youtube recommendations. In Proceedings of the 10th ACM Conference on Recommender Systems, RecSys ’16, page 191–198. find
  9. Zeyu Cui, Jianxin Ma, Chang Zhou, Jingren Zhou, and Hongxia Yang. 2022. M6-rec: Generative pretrained language models are open-ended recommender systems. arXiv preprint arXiv:2205.08084. — [arXiv:2205.08084](https://arxiv.org/abs/2205.08084) find
  10. W Dai, J Li, D Li, AMH Tiong, J Zhao, W Wang, B Li, P Fung, and S Hoi. 2023. Instructblip: Towards general-purpose vision-language models with instruction tuning. arXiv preprint arXiv:2305.06500.
  11. Gabriel de Souza Pereira Moreira, Sara Rabhi, Jeong Min Lee, Ronay Ak, and Even Oldridge. 2021. Transformers4rec: Bridging the gap between nlp and sequential/session-based recommendation. In Fifteenth ACM Conference on Recommender Systems, pages 143–153. find
  12. Li Dong, Shaohan Huang, Furu Wei, Mirella Lapata, Ming Zhou, and Ke Xu. 2017. Learning to generate product reviews from attributes. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 623–632. find
  13. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, and 1 others. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  14. Xinyan Fan, Zheng Liu, Jianxun Lian, Wayne Xin Zhao, Xing Xie, and Ji-Rong Wen. 2021. Lighter and better: low-rank decomposed self-attention networks for next-item recommendation. In Proceedings of the 44th international ACM SIGIR conference on research and development in information retrieval, page… find
  15. Shijie Geng, Zuohui Fu, Yingqiang Ge, Lei Li, Gerard de Melo, and Yongfeng Zhang. 2022a. Improving personalized explanation generation through visualization. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 244–255. find
  16. Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022b. Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5). In Proceedings of the 16th ACM Conference on Recommender Systems, pages 299–315. find
  17. Shijie Geng, Juntao Tan, Shuchang Liu, Zuohui Fu, and Yongfeng Zhang. 2023. Vip5: Towards multimodal foundation models for recommendation. arXiv preprint arXiv:2305.14302. find
  18. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948.
  19. Ruining He and Julian McAuley. 2016. Vbpr: visual bayesian personalized ranking from implicit feedback. In Proceedings of the AAAI conference on artificial intelligence, volume 30. find
  20. Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, pages 639–648. find
  21. Min Hou, Le Wu, Enhong Chen, Zhi Li, Vincent W Zheng, and Qi Liu. 2019. Explainable fashion recommendation: A semantic attribute region guided approach. arXiv preprint arXiv:1905.12862. — [arXiv:1905.12862](https://arxiv.org/abs/1905.12862) find
  22. Yupeng Hou, Shanlei Mu, Wayne Xin Zhao, Yaliang Li, Bolin Ding, and Ji-Rong Wen. 2022. Towards universal sequence representation learning for recommender systems. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 585–593. find
  23. Wenyue Hua, Shuyuan Xu, Yingqiang Ge, and Yongfeng Zhang. 2023. How to Index Item IDs for Recommendation Foundation Models. In Proceedings of 1st International ACM SIGIR Conference on Information Retrieval in the Asia Pacific (SIGIR-AP). find
  24. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  25. Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recommendation. In 2018 IEEE international conference on data mining (ICDM), pages 197–206. IEEE. find
  26. Bo Li, Yuanhan Zhang, Liangyu Chen, Jinghao Wang, Fanyi Pu, Jingkang Yang, Chunyuan Li, and Ziwei Liu. 2023a. Mimic-it: Multi-modal in-context instruction tuning. arXiv preprint arXiv:2306.05425. — [arXiv:2306.05425](https://arxiv.org/abs/2306.05425) find
  27. Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023b. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. arXiv preprint arXiv:2301.12597.
  28. Lei Li, Yongfeng Zhang, and Li Chen. 2021. Personalized transformer for explainable recommendation. arXiv preprint arXiv:2105.11601. find
  29. Lei Li, Yongfeng Zhang, and Li Chen. 2022. Personalized prompt learning for explainable recommendation. ACM Transactions on Information Systems. find
  30. Lei Li, Yongfeng Zhang, Dugang Liu, and Li Chen. 2023c. Large language models for generative recommendation: A survey and visionary discussions. arXiv:2309.01157. find
  31. Tianle Li, Wei-Lin Chiang, Evan Frick, Lisa Dunlap, Tianhao Wu, Banghua Zhu, Joseph E Gonzalez, and Ion Stoica. 2024. From crowdsourced data to high-quality benchmarks: Arena-hard and benchbuilder pipeline. arXiv preprint arXiv:2406.11939.
  32. Jianxun Lian, Xiaohuan Zhou, Fuzheng Zhang, Zhongxia Chen, Xing Xie, and Guangzhong Sun. 2018. xdeepfm: Combining explicit and implicit feature interactions for recommender systems. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining, pages 1754–1763. find
  33. Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, and 1 others. 2024a. Deepseek-v3 technical report. arXiv preprint arXiv:2412.19437. find
  34. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2023a. Improved baselines with visual instruction tuning. arXiv preprint arXiv:2310.03744.
  35. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024b. Improved baselines with visual instruction tuning. In CVPR.
  36. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023b. Visual instruction tuning. arXiv preprint arXiv:2304.08485.
  37. Shilong Liu, Hao Cheng, Haotian Liu, Hao Zhang, Feng Li, Tianhe Ren, Xueyan Zou, Jianwei Yang, Hang Su, Jun Zhu, and 1 others. 2023c. Llava-plus: Learning to use tools for creating multimodal agents. arXiv preprint arXiv:2311.05437. — [arXiv:2311.05437](https://arxiv.org/abs/2311.05437) find
  38. MAA. 2024. American invitational mathematics examination - aime. In American Invitational Mathematics Examination - AIME 2024. find
  39. Kelong Mao, Jieming Zhu, Liangcai Su, Guohao Cai, Yuru Li, and Zhenhua Dong. 2023. Finalmlp: An enhanced two-stream mlp model for ctr prediction. arXiv preprint arXiv:2304.00902. — [arXiv:2304.00902](https://arxiv.org/abs/2304.00902) find
  40. Lei Meng, Fuli Feng, Xiangnan He, Xiaoyan Gao, and Tat-Seng Chua. 2020. Heterogeneous fusion of semantic and collaborative information for visually-aware food recommendation. In Proceedings of the 28th ACM International Conference on Multimedia, pages 3460–3468. find
  41. Maxim Naumov, Dheevatsa Mudigere, Hao-Jun Michael Shi, Jianyu Huang, Narayanan Sundaraman, Jongsoo Park, Xiaodong Wang, Udit Gupta, Carole-Jean Wu, Alisson G Azzolini, and 1 others. 2019. Deep learning recommendation model for personalization and recommendation systems. arXiv preprint arXiv:1906.000… — [arXiv:1906.00091](https://arxiv.org/abs/1906.00091) find
  42. Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2012. Bpr: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618. — [arXiv:1205.2618](https://arxiv.org/abs/1205.2618) find
  43. Weiping Song, Chence Shi, Zhiping Xiao, Zhijian Duan, Yewen Xu, Ming Zhang, and Jian Tang. 2019. Autoint: Automatic feature interaction learning via self-attentive neural networks. In Proceedings of the 28th ACM international conference on information and knowledge management, pages 1161–1170. find
  44. Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang. 2019. Bert4rec: Sequential recommendation with bidirectional encoder representations from transformer. In Proceedings of the 28th ACM international conference on information and knowledge management, pages 1441–1450. find
  45. Zhiqing Sun, Sheng Shen, Shengcao Cao, Haotian Liu, Chunyuan Li, Yikang Shen, Chuang Gan, Liang-Yan Gui, Yu-Xiong Wang, Yiming Yang, and 1 others. 2023. Aligning large multimodal models with factually augmented rlhf. arXiv preprint arXiv:2309.14525.
  46. Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, and 1 others. 2023a. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805.
  47. Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, and 1 others. 2023b. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805.
  48. Dhruv Verma, Kshitij Gulati, Vasu Goel, and Rajiv Ratn Shah. 2020. Fashionist: Personalising outfit recommendation for cold-start scenarios. In Proceedings of the 28th ACM International Conference on Multimedia, pages 4527–4529. find
  49. Jie Wang, Fajie Yuan, Mingyue Cheng, Joemon M Jose, Chenyun Yu, Beibei Kong, Zhijin Wang, Bo Hu, and Zang Li. 2022. Transrec: Learning transferable recommendation from mixture-of-modality feedback. arXiv preprint arXiv:2206.06190. — [arXiv:2206.06190](https://arxiv.org/abs/2206.06190) find
  50. Ruoxi Wang, Rakesh Shivanna, Derek Cheng, Sagar Jain, Dong Lin, Lichan Hong, and Ed Chi. 2021a. Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems. In Proceedings of the Web Conference 2021, WWW ’21, page 1785–1797, New York, NY, USA. Association for C… find
  51. Ruoxi Wang, Rakesh Shivanna, Derek Cheng, Sagar Jain, Dong Lin, Lichan Hong, and Ed Chi. 2021b. Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems. In Proceedings of the web conference 2021, pages 1785–1797. find
  52. Wenhai Wang, Zhe Chen, Xiaokang Chen, Jiannan Wu, Xizhou Zhu, Gang Zeng, Ping Luo, Tong Lu, Jie Zhou, Yu Qiao, and 1 others. 2023. Visionllm: Large language model is also an open-ended decoder for vision-centric tasks. NeurlPS. find
  53. Zhiqiang Wang, Qingyun She, and Junlin Zhang. 2021c. Masknet: Introducing feature-wise multiplication to ctr ranking models by instance-guided mask. arXiv preprint arXiv:2102.07619. — [arXiv:2102.07619](https://arxiv.org/abs/2102.07619) find
  54. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, and 1 others. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824– 24837.
  55. Tianxin Wei, Bowen Jin, Ruirui Li, Hansi Zeng, Zhengyang Wang, Jianhui Sun, Qingyu Yin, Hanqing Lu, Suhang Wang, Jingrui He, and 1 others. 2024. Towards unified multi-modal personalization: Large vision-language models for generative recommendation and beyond. arXiv preprint arXiv:2403.10667. — [arXiv:2403.10667](https://arxiv.org/abs/2403.10667) find
  56. Jian Xie, Kai Zhang, Jiangjie Chen, Tinghui Zhu, Renze Lou, Yuandong Tian, Yanghua Xiao, and Yu Su. 2024a. Travelplanner: A benchmark for real-world planning with language agents. arXiv preprint arXiv:2402.01622.
  57. Junlin Xie, Zhihong Chen, Ruifei Zhang, Xiang Wan, and Guanbin Li. 2024b. Large multimodal agents: A survey. arXiv preprint arXiv:2402.15116. — [arXiv:2402.15116](https://arxiv.org/abs/2402.15116) find
  58. Shuyuan Xu, Wenyue Hua, and Yongfeng Zhang. 2023. OpenP5: Benchmarking Foundation Models for Recommendation. arXiv:2306.11134. — [arXiv:2306.11134](https://arxiv.org/abs/2306.11134) find
  59. Jieming Zhu, Quanyu Dai, Liangcai Su, Rong Ma, Jinyang Liu, Guohao Cai, Xi Xiao, and Rui Zhang. 2022. BARS: towards open benchmarking for recommender systems. In SIGIR ’22: The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, Madrid, Spain, July 11 - 15,… find
  60. Jieming Zhu, Jinyang Liu, Shuai Yang, Qi Zhang, and Xiuqiang He. 2021. Open benchmarking for clickthrough rate prediction. In CIKM ’21: The 30th ACM International Conference on Information and Knowledge Management, Virtual Event, Queensland, Australia, November 1 - 5, 2021, pages 2759–2769. ACM. find
  61. Yongshuo Zong, Ondrej Bohdal, Tingyang Yu, Yongxin Yang, and Timothy Hospedales. 2024. Safety fine-tuning at (almost) no cost: a baseline for vision large language models. In ICML.
  62. Qinghao Ye, Haiyang Xu, Guohai Xu, Jiabo Ye, Ming Yan, Yiyang Zhou, Junyang Wang, Anwen Hu, Pengcheng Shi, Yaya Shi, and 1 others. 2023. mplug-owl: Modularization empowers large language models with multimodality. arXiv preprint arXiv:2304.14178.
  63. Fajie Yuan, Alexandros Karatzoglou, Ioannis Arapakis, Joemon M Jose, and Xiangnan He. 2019. A simple convolutional generative network for next item recommendation. In Proceedings of the twelfth ACM international conference on web search and data mining, pages 582–590. find
  64. Buyun Zhang, Liang Luo, Yuxin Chen, Jade Nie, Xi Liu, Daifeng Guo, Yanli Zhao, Shen Li, Yuchen Hao, Yantao Yao, and 1 others. 2024. Wukong: Towards a scaling law for large-scale recommendation. arXiv preprint arXiv:2403.02545. — [arXiv:2403.02545](https://arxiv.org/abs/2403.02545) find
  65. Yongfeng Zhang, Qingyao Ai, Xu Chen, and W Bruce Croft. 2017. Joint representation learning for topn recommendation with heterogeneous information sources. In Proceedings of the 2017 ACM on Conference on Information and Knowledge Management, pages 1449–1458. find
  66. Kun Zhou, Hui Wang, Wayne Xin Zhao, Yutao Zhu, Sirui Wang, Fuzheng Zhang, Zhongyuan Wang, and Ji-Rong Wen. 2020. S3-rec: Self-supervised learning for sequential recommendation with mutual information maximization. In Proceedings of the 29th ACM international conference on information & knowledge man… find
  67. Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. 2023. Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592.
  68. Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. 2023. MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models. arXiv preprint, page arXiv:2304.10592.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhang-2026-reasonrec-reasoning-augmented-multimodal-agent/