---
title: "Multimodal LLMs Do Not Compose Skills — Ontalvilla 2026"
description: "The paper asks whether multimodal LLMs can chain a visual ability with a textual one to solve tasks they were never directly trained on."
authors: ["Paula Ontalvilla", "Aitor Ormazabal", "Gorka Azkune"]
year: 2026
venue: "Proceedings of the Fifteenth Language Resources and Evaluation Conference"
source: "https://aclanthology.org/2026.lrec-1.767/"
doi: "10.63317/32osokhu3sac"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/ontalvilla-2026-multimodal-llms-do-not-compose/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ontalvilla-2026-multimodal-llms-do-not-compose/)

[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/7/) / Multimodal LLMs Do Not Compose Skills Optimally across Modalities

# Multimodal LLMs Do Not Compose Skills Optimally across Modalities

**Authors:** Paula Ontalvilla, Aitor Ormazabal, Gorka Azkune · **Year:** 2026 · **Venue:** Proceedings of the Fifteenth Language Resources and Evaluation Conference · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.lrec-1.767/) · [PDF](https://aclanthology.org/2026.lrec-1.767.pdf) · [DOI 10.63317/32osokhu3sac](https://doi.org/10.63317/32osokhu3sac) · [OpenAlex](https://openalex.org/W4416233064)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14)

## Summary

The paper asks whether multimodal LLMs can chain a visual ability with a textual one to solve tasks they were never directly trained on. The authors design three image-to-text tasks. Each can be solved with one visual step followed by one textual step:

- solving problems shown as rendered images (OCR plus reasoning),
- counting objects (detection plus counting),
- reasoning about four poker cards (recognition plus numerical reasoning).

Open-weight late-fusion MLLMs are tested in two modes. In the first, they answer directly. In the second, a cascade is used: the same model first outputs an intermediate text, such as a transcript, bounding boxes or a card list, and is then queried again to finish the task. The performance difference between the two modes defines a cross-modality skill composition gap.

Every model shows a substantial gap, even though the compositions are trivial for humans. Task-specific chain-of-thought prompts and a dedicated fine-tuning recipe both narrow the gap, but neither closes it. The prompting approach also scales poorly, since prompts must be written per task. The work exposes a basic weakness in how MLLMs integrate modalities and provides a reusable evaluation setup.

## Key points

- Evaluated models: LLaVA 1.6-Vicuna-13B, LLaVA 1.6-Mistral-7B, Llama 3.2 11B, Molmo-7B and 72B, and Qwen2.5-VL-7B and 72B.
- Closed models were excluded because their training skills cannot be verified.
- LLaVA and Llama were not explicitly trained for object detection.
- The appendix examples draw on GSM8K (rendered-text reasoning) and CV-Bench (object counting).
- Five CoT prompt variants (P1–P5) are listed for the card task.
- Fine-tuning improves results and partly generalizes, but a gap remains in all settings.
- Code is available at github.com/paulaonta/skillComposition. The paper appears in LREC 2026, pp. 9767–9789.

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

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cite

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

```bibtex
@inproceedings{ontalvilla2026multimodal,
  title = {{Multimodal LLMs Do Not Compose Skills Optimally across Modalities}},
  author = {Ontalvilla, Paula and Ormazabal, Aitor and Azkune, Gorka},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/32osokhu3sac},
  url = {https://doi.org/10.63317/32osokhu3sac}
}
```

- [← previous Multimodal Entrainment and Feedback in Online Group Meetings](https://md.research.elamur.ai/free/en/papers/paggio-2026-multimodal-entrainment-feedback-online-group/)
- [next → Multimodal Reference by Means of the Pronoun We and Hand Gestures in a Novel Corpus of Parliamentary Opening Debates](https://md.research.elamur.ai/free/en/papers/navarretta-2026-multimodal-reference-means-pronoun-we/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/7/)

## References

42 references, 20 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. 2023. Gpt-4v(ision) system card. [find](https://scholar.google.com/scholar?q=2023%2E+Gpt%2D4v%28ision%29+system+card%2E)
2. Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katie Millican, Malcolm Reynolds, Roman Ring, Eliza Rutherford, Serkan Cabi, Tengda Han, Zhitao Gong, Sina Samangooei, Marianne Monteiro, Jacob Menick, Sebastian Borgeaud, Andrew Brock… [find](https://scholar.google.com/scholar?q=Jean%2DBaptiste+Alayrac%2C+Jeff+Donahue%2C+Pauline+Luc%2C+Antoine+Miech%2C+Iain+Barr%2C+Yana+Hasson%2C+Karel+Lenc%2C+Arthur+Mensch%2C+Katie+Millican%2C+Malcolm+Reynolds%2C+Roman+Ring%2C+Eliza+Rutherford%2C+Serkan+Cabi%2C+Tengda+) [our card](https://md.research.elamur.ai/free/en/papers/alayrac-2022-flamingo-visual-language-model-few/)
3. Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al. 2025. Qwen2. 5-vl technical report. arXiv preprint arXiv:2502.13923. [find](https://scholar.google.com/scholar?q=Shuai+Bai%2C+Keqin+Chen%2C+Xuejing+Liu%2C+Jialin+Wang%2C+Wenbin+Ge%2C+Sibo+Song%2C+Kai+Dang%2C+Peng+Wang%2C+Shijie+Wang%2C+Jun+Tang%2C+et+al%2E+2025%2E+Qwen2%2E+5%2Dvl+technical+report%2E+arXiv+preprint+arXiv%3A2502%2E13923%2E) [our card](https://md.research.elamur.ai/free/en/papers/bai-2025-qwen2-5-vl-technical/)
4. Tianzhe Chu, Yuexiang Zhai, Jihan Yang, Shengbang Tong, Saining Xie, Dale Schuurmans, Quoc V. Le, Sergey Levine, and Yi Ma. 2025. Sft memorizes, rl generalizes: A comparative study of foundation model post-training. [find](https://scholar.google.com/scholar?q=Tianzhe+Chu%2C+Yuexiang+Zhai%2C+Jihan+Yang%2C+Shengbang+Tong%2C+Saining+Xie%2C+Dale+Schuurmans%2C+Quoc+V%2E+Le%2C+Sergey+Levine%2C+and+Yi+Ma%2E+2025%2E+Sft+memorizes%2C+rl+generalizes%3A+A+comparative+study+of+foundation+model) [our card](https://md.research.elamur.ai/free/en/papers/chu-2025-sft-memorizes-rl-generalizes-comparative/)
5. Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. 2018. Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457. [find](https://scholar.google.com/scholar?q=Peter+Clark%2C+Isaac+Cowhey%2C+Oren+Etzioni%2C+Tushar+Khot%2C+Ashish+Sabharwal%2C+Carissa+Schoenick%2C+and+Oyvind+Tafjord%2E+2018%2E+Think+you+have+solved+question+answering%3F+try+arc%2C+the+ai2+reasoning+challenge%2E+arX)
6. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168. [find](https://scholar.google.com/scholar?q=Karl+Cobbe%2C+Vineet+Kosaraju%2C+Mohammad+Bavarian%2C+Mark+Chen%2C+Heewoo+Jun%2C+Lukasz+Kaiser%2C+Matthias+Plappert%2C+Jerry+Tworek%2C+Jacob+Hilton%2C+Reiichiro+Nakano%2C+Christopher+Hesse%2C+and+John+Schulman%2E+2021%2E+Train) [our card](https://md.research.elamur.ai/free/en/papers/cobbe-2021-training-verifiers-solve-math-word/)
7. Wenliang Dai, Nayeon Lee, Boxin Wang, Zhuolin Yang, Zihan Liu, Jon Barker, Tuomas Rintamaki, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. 2024. Nvlm: Open frontier-class multimodal llms. [find](https://scholar.google.com/scholar?q=Wenliang+Dai%2C+Nayeon+Lee%2C+Boxin+Wang%2C+Zhuolin+Yang%2C+Zihan+Liu%2C+Jon+Barker%2C+Tuomas+Rintamaki%2C+Mohammad+Shoeybi%2C+Bryan+Catanzaro%2C+and+Wei+Ping%2E+2024%2E+Nvlm%3A+Open+frontier%2Dclass+multimodal+llms%2E)
8. Yunkai Dang, Kaichen Huang, Jiahao Huo, Yibo Yan, Sirui Huang, Dongrui Liu, Mengxi Gao, Jie Zhang, Chen Qian, Kun Wang, Yong Liu, Jing Shao, Hui Xiong, and Xuming Hu. 2024. Explainable and interpretable multimodal large language models: A comprehensive survey. ArXiv, abs/2412.02104. — [arXiv:2412.02104](https://arxiv.org/abs/2412.02104) [find](https://scholar.google.com/scholar?q=Yunkai+Dang%2C+Kaichen+Huang%2C+Jiahao+Huo%2C+Yibo+Yan%2C+Sirui+Huang%2C+Dongrui+Liu%2C+Mengxi+Gao%2C+Jie+Zhang%2C+Chen+Qian%2C+Kun+Wang%2C+Yong+Liu%2C+Jing+Shao%2C+Hui+Xiong%2C+and+Xuming+Hu%2E+2024%2E+Explainable+and+interpretab)
9. DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, Z. F. Wu, Zhibin Gou, Zhihong Shao, Zhuoshu Li, Ziyi Gao, Aixin Liu, Bing Xue, Bingxuan Wang, Bochao Wu, Bei Feng, Chengda Lu, Chengg… [find](https://scholar.google.com/scholar?q=DeepSeek%2DAI%2C+Daya+Guo%2C+Dejian+Yang%2C+Haowei+Zhang%2C+Junxiao+Song%2C+Ruoyu+Zhang%2C+Runxin+Xu%2C+Qihao+Zhu%2C+Shirong+Ma%2C+Peiyi+Wang%2C+Xiao+Bi%2C+Xiaokang+Zhang%2C+Xingkai+Yu%2C+Yu+Wu%2C+Z%2E+F%2E+Wu%2C+Zhibin+Gou%2C+Zhihong+Sha) [our card](https://md.research.elamur.ai/free/en/papers/deepseek-2025-ai-deepseek-r1-incentivizing-reasoning-capability/)
10. Matt Deitke, Christopher Clark, Sangho Lee, Rohun Tripathi, Yue Yang, Jae Sung Park, Mohammadreza Salehi, Niklas Muennighoff, Kyle Lo, Luca Soldaini, et al. 2024. Molmo and pixmo: Open weights and open data for state-of-the-art multimodal models. CoRR. [find](https://scholar.google.com/scholar?q=Matt+Deitke%2C+Christopher+Clark%2C+Sangho+Lee%2C+Rohun+Tripathi%2C+Yue+Yang%2C+Jae+Sung+Park%2C+Mohammadreza+Salehi%2C+Niklas+Muennighoff%2C+Kyle+Lo%2C+Luca+Soldaini%2C+et+al%2E+2024%2E+Molmo+and+pixmo%3A+Open+weights+and+ope)
11. Danny Driess, Fei Xia, Mehdi S. M. Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, Yevgen Chebotar, Pierre Sermanet, Daniel Duckworth, Sergey Levine, Vincent Vanhoucke, Karol Hausman, Marc Toussaint, Klaus Greff, Andy Zen… [find](https://scholar.google.com/scholar?q=Danny+Driess%2C+Fei+Xia%2C+Mehdi+S%2E+M%2E+Sajjadi%2C+Corey+Lynch%2C+Aakanksha+Chowdhery%2C+Brian+Ichter%2C+Ayzaan+Wahid%2C+Jonathan+Tompson%2C+Quan+Vuong%2C+Tianhe+Yu%2C+Wenlong+Huang%2C+Yevgen+Chebotar%2C+Pierre+Sermanet%2C+Dani)
12. Leo Gao, Jonathan Tow, Baber Abbasi, Stella Biderman, Sid Black, Anthony DiPofi, Charles Foster, Laurence Golding, Jeffrey Hsu, Alain Le Noac’h, Haonan Li, Kyle McDonell, Niklas Muennighoff, Chris Ociepa, Jason Phang, Laria Reynolds, Hailey Schoelkopf, Aviya Skowron, Lintang Sutawika, Eric Tang, Ani… [find](https://scholar.google.com/scholar?q=Leo+Gao%2C+Jonathan+Tow%2C+Baber+Abbasi%2C+Stella+Biderman%2C+Sid+Black%2C+Anthony+DiPofi%2C+Charles+Foster%2C+Laurence+Golding%2C+Jeffrey+Hsu%2C+Alain+Le+Noac%E2%80%99h%2C+Haonan+Li%2C+Kyle+McDonell%2C+Niklas+Muennighoff%2C+Chris+Oci)
13. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783. [find](https://scholar.google.com/scholar?q=Aaron+Grattafiori%2C+Abhimanyu+Dubey%2C+Abhinav+Jauhri%2C+Abhinav+Pandey%2C+Abhishek+Kadian%2C+Ahmad+Al%2DDahle%2C+Aiesha+Letman%2C+Akhil+Mathur%2C+Alan+Schelten%2C+Alex+Vaughan%2C+et+al%2E+2024%2E+The+llama+3+herd+of+models%2E+)
14. Tianyu He, Darshil Doshi, Aritra Das, and Andrey Gromov. 2024. Learning to grok: Emergence of in-context learning and skill composition in modular arithmetic tasks. Advances in Neural Information Processing Systems, 37:13244–13273. [find](https://scholar.google.com/scholar?q=Tianyu+He%2C+Darshil+Doshi%2C+Aritra+Das%2C+and+Andrey+Gromov%2E+2024%2E+Learning+to+grok%3A+Emergence+of+in%2Dcontext+learning+and+skill+composition+in+modular+arithmetic+tasks%2E+Advances+in+Neural+Information+Proc)
15. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. Measuring massive multi-task language understanding. arXiv preprint arXiv:2009.03300. [find](https://scholar.google.com/scholar?q=Dan+Hendrycks%2C+Collin+Burns%2C+Steven+Basart%2C+Andy+Zou%2C+Mantas+Mazeika%2C+Dawn+Song%2C+and+Jacob+Steinhardt%2E+2020%2E+Measuring+massive+multi%2Dtask+language+understanding%2E+arXiv+preprint+arXiv%3A2009%2E03300%2E) [our card](https://md.research.elamur.ai/free/en/papers/hendrycks-2020-measuring-massive-multitask-language-understanding/)
16. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2024. Measuring mathematical problem solving with the math dataset, 2021. URL https://arxiv.org/abs/2103.03874. [↗](https://arxiv.org/abs/2103.03874) [our card](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-measuring-mathematical-problem-solving-math/)
17. Brenden M Lake and Marco Baroni. 2023. Human-like systematic generalization through a meta-learning neural network. Nature, 623(7985):115– 121. [find](https://scholar.google.com/scholar?q=Brenden+M+Lake+and+Marco+Baroni%2E+2023%2E+Human%2Dlike+systematic+generalization+through+a+meta%2Dlearning+neural+network%2E+Nature%2C+623%287985%29%3A115%E2%80%93+121%2E)
18. Xiujun Li, Yujie Lu, William Yang Wang, and Yejin Choi. 2024. Pixelated instructions: Can multimodal large language models follow printed instructions in images? In Adaptive Foundation Models: Evolving AI for Personalized and Efficient Learning. [find](https://scholar.google.com/scholar?q=Xiujun+Li%2C+Yujie+Lu%2C+William+Yang+Wang%2C+and+Yejin+Choi%2E+2024%2E+Pixelated+instructions%3A+Can+multimodal+large+language+models+follow+printed+instructions+in+images%3F+In+Adaptive+Foundation+Models%3A+Evolvin)
19. Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. 2023. Let’s verify step by step. In The Twelfth International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Hunter+Lightman%2C+Vineet+Kosaraju%2C+Yuri+Burda%2C+Harrison+Edwards%2C+Bowen+Baker%2C+Teddy+Lee%2C+Jan+Leike%2C+John+Schulman%2C+Ilya+Sutskever%2C+and+Karl+Cobbe%2E+2023%2E+Let%E2%80%99s+verify+step+by+step%2E+In+The+Twelfth+Intern) [our card](https://md.research.elamur.ai/free/en/papers/lightman-2023-let-s-verify-step-step/)
20. Ji Lin, Hongxu Yin, Wei Ping, Yao Lu, Pavlo Molchanov, Andrew Tao, Huizi Mao, Jan Kautz, Mohammad Shoeybi, and Song Han. 2023. Vila: On pre-training for visual language models. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 26679–26689. [find](https://scholar.google.com/scholar?q=Ji+Lin%2C+Hongxu+Yin%2C+Wei+Ping%2C+Yao+Lu%2C+Pavlo+Molchanov%2C+Andrew+Tao%2C+Huizi+Mao%2C+Jan+Kautz%2C+Mohammad+Shoeybi%2C+and+Song+Han%2E+2023%2E+Vila%3A+On+pre%2Dtraining+for+visual+language+models%2E+2024+IEEE%2FCVF+Conferenc) [our card](https://md.research.elamur.ai/free/en/papers/lin-2023-vila-pre-training-visual-language/)
21. Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. In Computer vision–ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedings, part v 13, page… [find](https://scholar.google.com/scholar?q=Tsung%2DYi+Lin%2C+Michael+Maire%2C+Serge+Belongie%2C+James+Hays%2C+Pietro+Perona%2C+Deva+Ramanan%2C+Piotr+Doll%C3%A1r%2C+and+C+Lawrence+Zitnick%2E+2014%2E+Microsoft+coco%3A+Common+objects+in+context%2E+In+Computer+vision%E2%80%93ECCV+201) [our card](https://md.research.elamur.ai/free/en/papers/lin-2014-microsoft-coco-common-objects-context/)
22. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024a. Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306. [find](https://scholar.google.com/scholar?q=Haotian+Liu%2C+Chunyuan+Li%2C+Yuheng+Li%2C+and+Yong+Jae+Lee%2E+2024a%2E+Improved+baselines+with+visual+instruction+tuning%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+Computer+Vision+and+Pattern+Recognition%2C+p) [our card](https://md.research.elamur.ai/free/en/papers/liu-2023-improved-baselines-visual-instruction-tuning/)
23. Haotian Liu, Chunyuan Li, Yuheng Li, Bo Li, Yuanhan Zhang, Sheng Shen, and Yong Jae Lee. 2024b. Llava-next: Improved reasoning, ocr, and world knowledge. [find](https://scholar.google.com/scholar?q=Haotian+Liu%2C+Chunyuan+Li%2C+Yuheng+Li%2C+Bo+Li%2C+Yuanhan+Zhang%2C+Sheng+Shen%2C+and+Yong+Jae+Lee%2E+2024b%2E+Llava%2Dnext%3A+Improved+reasoning%2C+ocr%2C+and+world+knowledge%2E)
24. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. Visual instruction tuning. [find](https://scholar.google.com/scholar?q=Haotian+Liu%2C+Chunyuan+Li%2C+Qingyang+Wu%2C+and+Yong+Jae+Lee%2E+2023%2E+Visual+instruction+tuning%2E) [our card](https://md.research.elamur.ai/free/en/papers/liu-2023-visual-instruction-tuning/)
25. Tenglong Liu, Jianxiong Li, Yinan Zheng, Haoyi Niu, Yixing Lan, Xin Xu, and Xianyuan Zhan. 2025. Skill expansion and composition in parameter space. arXiv preprint arXiv:2502.05932. — [arXiv:2502.05932](https://arxiv.org/abs/2502.05932) [find](https://scholar.google.com/scholar?q=Tenglong+Liu%2C+Jianxiong+Li%2C+Yinan+Zheng%2C+Haoyi+Niu%2C+Yixing+Lan%2C+Xin+Xu%2C+and+Xianyuan+Zhan%2E+2025%2E+Skill+expansion+and+composition+in+parameter+space%2E+arXiv+preprint+arXiv%3A2502%2E05932%2E+%E2%80%94+%5BarXiv%3A2502%2E0593)
26. Pan Lu, Baolin Peng, Hao Cheng, Michel Galley, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, and Jianfeng Gao. 2023. Chameleon: Plug-and-play compositional reasoning with large language models. In Advances in Neural Information Processing Systems, volume 36, pages 43447– 43478. Curran Associates, Inc. [find](https://scholar.google.com/scholar?q=Pan+Lu%2C+Baolin+Peng%2C+Hao+Cheng%2C+Michel+Galley%2C+Kai%2DWei+Chang%2C+Ying+Nian+Wu%2C+Song%2DChun+Zhu%2C+and+Jianfeng+Gao%2E+2023%2E+Chameleon%3A+Plug%2Dand%2Dplay+compositional+reasoning+with+large+language+models%2E+In+Advan) [our card](https://md.research.elamur.ai/free/en/papers/lu-2023-chameleon-plug-play-compositional-reasoning/)
27. OpenAI, :, Aaron Jaech, Adam Kalai, Adam Lerer, Adam Richardson, Ahmed El-Kishky, Aiden Low, Alec Helyar, Aleksander Madry, Alex Beutel, Alex Carney, Alex Iftimie, Alex Karpenko, Alex Tachard Passos, Alexander Neitz, Alexander Prokofiev, Alexander Wei, Allison Tam, Ally Bennett, Ananya Kumar, Andre… [find](https://scholar.google.com/scholar?q=OpenAI%2C+%3A%2C+Aaron+Jaech%2C+Adam+Kalai%2C+Adam+Lerer%2C+Adam+Richardson%2C+Ahmed+El%2DKishky%2C+Aiden+Low%2C+Alec+Helyar%2C+Aleksander+Madry%2C+Alex+Beutel%2C+Alex+Carney%2C+Alex+Iftimie%2C+Alex+Karpenko%2C+Alex+Tachard+Passos%2C+) [our card](https://md.research.elamur.ai/free/en/papers/openai-2024-openai-o1-system-card/)
28. OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mohammad Bavarian, Jeff Belgum, Irwan Bello, Ja… [find](https://scholar.google.com/scholar?q=OpenAI%2C+Josh+Achiam%2C+Steven+Adler%2C+Sandhini+Agarwal%2C+Lama+Ahmad%2C+Ilge+Akkaya%2C+Florencia+Leoni+Aleman%2C+Diogo+Almeida%2C+Janko+Altenschmidt%2C+Sam+Altman%2C+Shyamal+Anadkat%2C+Red+Avila%2C+Igor+Babuschkin%2C+Suchir) [our card](https://md.research.elamur.ai/free/en/papers/openai-2023-gpt-4-technical-report/)
29. Ahmed H Qureshi, Jacob J Johnson, Yuzhe Qin, Taylor Henderson, Byron Boots, and Michael C Yip. Composing task-agnostic policies with deep reinforcement learning. In International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Ahmed+H+Qureshi%2C+Jacob+J+Johnson%2C+Yuzhe+Qin%2C+Taylor+Henderson%2C+Byron+Boots%2C+and+Michael+C+Yip%2E+Composing+task%2Dagnostic+policies+with+deep+reinforcement+learning%2E+In+International+Conference+on+Learnin)
30. Haozhan Shen, Peng Liu, Jingcheng Li, Chunxin Fang, Yibo Ma, Jiajia Liao, Qiaoli Shen, Zilun Zhang, Kangjia Zhao, Qianqian Zhang, Ruochen Xu, and Tiancheng Zhao. 2025. Vlm-r1: A stable and generalizable r1-style large vision-language model. [find](https://scholar.google.com/scholar?q=Haozhan+Shen%2C+Peng+Liu%2C+Jingcheng+Li%2C+Chunxin+Fang%2C+Yibo+Ma%2C+Jiajia+Liao%2C+Qiaoli+Shen%2C+Zilun+Zhang%2C+Kangjia+Zhao%2C+Qianqian+Zhang%2C+Ruochen+Xu%2C+and+Tiancheng+Zhao%2E+2025%2E+Vlm%2Dr1%3A+A+stable+and+generalizab)
31. Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805. [find](https://scholar.google.com/scholar?q=Gemini+Team%2C+Rohan+Anil%2C+Sebastian+Borgeaud%2C+Jean%2DBaptiste+Alayrac%2C+Jiahui+Yu%2C+Radu+Soricut%2C+Johan+Schalkwyk%2C+Andrew+M+Dai%2C+Anja+Hauth%2C+Katie+Millican%2C+et+al%2E+2023%2E+Gemini%3A+a+family+of+highly+capable+) [our card](https://md.research.elamur.ai/free/en/papers/gemini-2023-team-gemini-family-highly-capable-multimodal/)
32. Reka Team, Aitor Ormazabal, Che Zheng, Cyprien de Masson d’Autume, Dani Yogatama, Deyu Fu, Donovan Ong, Eric Chen, Eugenie Lamprecht, Hai Pham, et al. 2024. Reka core, flash, and edge: A series of powerful multimodal language models. arXiv preprint arXiv:2404.12387. [find](https://scholar.google.com/scholar?q=Reka+Team%2C+Aitor+Ormazabal%2C+Che+Zheng%2C+Cyprien+de+Masson+d%E2%80%99Autume%2C+Dani+Yogatama%2C+Deyu+Fu%2C+Donovan+Ong%2C+Eric+Chen%2C+Eugenie+Lamprecht%2C+Hai+Pham%2C+et+al%2E+2024%2E+Reka+core%2C+flash%2C+and+edge%3A+A+series+of+pow)
33. Peter Tong, Ellis Brown, Penghao Wu, Sanghyun Woo, Adithya Jairam Vedagiri IYER, Sai Charitha Akula, Shusheng Yang, Jihan Yang, Manoj Middepogu, Ziteng Wang, et al. 2024. Cambrian-1: A fully open, vision-centric exploration of multimodal llms. Advances in Neural Information Processing Systems, 37:87… [find](https://scholar.google.com/scholar?q=Peter+Tong%2C+Ellis+Brown%2C+Penghao+Wu%2C+Sanghyun+Woo%2C+Adithya+Jairam+Vedagiri+IYER%2C+Sai+Charitha+Akula%2C+Shusheng+Yang%2C+Jihan+Yang%2C+Manoj+Middepogu%2C+Ziteng+Wang%2C+et+al%2E+2024%2E+Cambrian%2D1%3A+A+fully+open%2C+vis) [our card](https://md.research.elamur.ai/free/en/papers/tong-2024-cambrian-1-fully-open-vision/)
34. Haozhe Wang, Chao Qu, Zuming Huang, Wei Chu, Fangzhen Lin, and Wenhu Chen. 2025a. Vl-rethinker: Incentivizing self-reflection of vision-language models with reinforcement learning. [find](https://scholar.google.com/scholar?q=Haozhe+Wang%2C+Chao+Qu%2C+Zuming+Huang%2C+Wei+Chu%2C+Fangzhen+Lin%2C+and+Wenhu+Chen%2E+2025a%2E+Vl%2Drethinker%3A+Incentivizing+self%2Dreflection+of+vision%2Dlanguage+models+with+reinforcement+learning%2E)
35. Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, Zhaokai Wang, Zhe Chen, Hongjie Zhang, Ganlin Yang, Haomin Wang, Qi Wei, Jinhui Yin, Wenhao Li, Erfei Cui, Guanzhou Chen, Zichen Ding, Changyao Tian, Zhenyu Wu, Jingjing Xie,… [find](https://scholar.google.com/scholar?q=Weiyun+Wang%2C+Zhangwei+Gao%2C+Lixin+Gu%2C+Hengjun+Pu%2C+Long+Cui%2C+Xingguang+Wei%2C+Zhaoyang+Liu%2C+Linglin+Jing%2C+Shenglong+Ye%2C+Jie+Shao%2C+Zhaokai+Wang%2C+Zhe+Chen%2C+Hongjie+Zhang%2C+Ganlin+Yang%2C+Haomin+Wang%2C+Qi+Wei%2C+J) [our card](https://md.research.elamur.ai/free/en/papers/wang-2025-internvl3-5-advancing-open/)
36. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824–24837. [find](https://scholar.google.com/scholar?q=Jason+Wei%2C+Xuezhi+Wang%2C+Dale+Schuurmans%2C+Maarten+Bosma%2C+Fei+Xia%2C+Ed+Chi%2C+Quoc+V+Le%2C+Denny+Zhou%2C+et+al%2E+2022%2E+Chain%2Dof%2Dthought+prompting+elicits+reasoning+in+large+language+models%2E+Advances+in+neural+i) [our card](https://md.research.elamur.ai/free/en/papers/wei-2022-chain-thought-prompting-elicits-reasoning/)
37. Dingli Yu, Simran Kaur, Arushi Gupta, Jonah Brown-Cohen, Anirudh Goyal, and Sanjeev Arora. 2023. Skill-mix: A flexible and expandable family of evaluations for ai models. In NeurIPS 2023 Workshop on Distribution Shifts: New Frontiers with Foundation Models. [find](https://scholar.google.com/scholar?q=Dingli+Yu%2C+Simran+Kaur%2C+Arushi+Gupta%2C+Jonah+Brown%2DCohen%2C+Anirudh+Goyal%2C+and+Sanjeev+Arora%2E+2023%2E+Skill%2Dmix%3A+A+flexible+and+expandable+family+of+evaluations+for+ai+models%2E+In+NeurIPS+2023+Workshop+on+D)
38. Xiang Yue, Tianyu Zheng, Yuansheng Ni, Yubo Wang, Kai Zhang, Shengbang Tong, Yuxuan Sun, Botao Yu, Ge Zhang, Huan Sun, et al. 2024. Mmmu-pro: A more robust multi-discipline multimodal understanding benchmark. CoRR. [find](https://scholar.google.com/scholar?q=Xiang+Yue%2C+Tianyu+Zheng%2C+Yuansheng+Ni%2C+Yubo+Wang%2C+Kai+Zhang%2C+Shengbang+Tong%2C+Yuxuan+Sun%2C+Botao+Yu%2C+Ge+Zhang%2C+Huan+Sun%2C+et+al%2E+2024%2E+Mmmu%2Dpro%3A+A+more+robust+multi%2Ddiscipline+multimodal+understanding+be)
39. Renrui Zhang, Dongzhi Jiang, Yichi Zhang, Haokun Lin, Ziyu Guo, Pengshuo Qiu, Aojun Zhou, Pan Lu, Kai-Wei Chang, Yu Qiao, et al. 2024. Mathverse: Does your multi-modal llm truly see the diagrams in visual math problems? In European Conference on Computer Vision, pages 169–186. Springer. [find](https://scholar.google.com/scholar?q=Renrui+Zhang%2C+Dongzhi+Jiang%2C+Yichi+Zhang%2C+Haokun+Lin%2C+Ziyu+Guo%2C+Pengshuo+Qiu%2C+Aojun+Zhou%2C+Pan+Lu%2C+Kai%2DWei+Chang%2C+Yu+Qiao%2C+et+al%2E+2024%2E+Mathverse%3A+Does+your+multi%2Dmodal+llm+truly+see+the+diagrams+in+vi) [our card](https://md.research.elamur.ai/free/en/papers/zhang-2024-mathverse-multi-modal-llm-truly/)
40. Haoyu Zhao, Simran Kaur, Dingli Yu, Anirudh Goyal, and Sanjeev Arora. 2024. Can models learn skill composition from examples? Advances in Neural Information Processing Systems, 37:102393–102427. [find](https://scholar.google.com/scholar?q=Haoyu+Zhao%2C+Simran+Kaur%2C+Dingli+Yu%2C+Anirudh+Goyal%2C+and+Sanjeev+Arora%2E+2024%2E+Can+models+learn+skill+composition+from+examples%3F+Advances+in+Neural+Information+Processing+Systems%2C+37%3A102393%E2%80%93102427%2E)
41. Boyuan Zheng, Michael Y. Fatemi, Xiaolong Jin, Zora Zhiruo Wang, Apurva Gandhi, Yueqi Song, Yu Gu, Jayanth Srinivasa, Gaowen Liu, Graham Neubig, and Yu Su. 2025. Skillweaver: Web agents can self-improve by discovering and honing skills. [find](https://scholar.google.com/scholar?q=Boyuan+Zheng%2C+Michael+Y%2E+Fatemi%2C+Xiaolong+Jin%2C+Zora+Zhiruo+Wang%2C+Apurva+Gandhi%2C+Yueqi+Song%2C+Yu+Gu%2C+Jayanth+Srinivasa%2C+Gaowen+Liu%2C+Graham+Neubig%2C+and+Yu+Su%2E+2025%2E+Skillweaver%3A+Web+agents+can+self%2Dimpro)
42. Yanli Zhou, Brenden M Lake, and Adina Williams. 2024. Compositional learning of functions in humans and machines. arXiv preprint arXiv:2403.12201. — [arXiv:2403.12201](https://arxiv.org/abs/2403.12201) [find](https://scholar.google.com/scholar?q=Yanli+Zhou%2C+Brenden+M+Lake%2C+and+Adina+Williams%2E+2024%2E+Compositional+learning+of+functions+in+humans+and+machines%2E+arXiv+preprint+arXiv%3A2403%2E12201%2E+%E2%80%94+%5BarXiv%3A2403%2E12201%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2403%2E12201)