---
title: "When Is Thinking Enough? Early Exit via Sufficiency Assessment"
description: "The authors propose DTSR (Dynamic Thought Sufficiency in Reasoning), a training-free decoding… 49 references, cited by 1 paper here."
authors: ["Yang Xiang", "Yixin Ji", "Ruotao Xu", "Dan Qiao", "Zheming Yang", "Juntao Li", "Min Zhang"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.1080/"
doi: "10.18653/v1/2026.acl-long.1080"
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/xiang-2026-when-is-thinking-enough-early/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/xiang-2026-when-is-thinking-enough-early/)

[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/19/) / When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning

# When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning

**Authors:** [Yang Xiang](https://md.research.elamur.ai/free/en/authors/yang-xiang/), Yixin Ji, Ruotao Xu, Dan Qiao, Zheming Yang, [Juntao Li](https://md.research.elamur.ai/free/en/authors/juntao-li/) and [Min Zhang](https://md.research.elamur.ai/free/en/authors/min-zhang/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/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.acl-long.1080/) · [PDF](https://aclanthology.org/2026.acl-long.1080.pdf) · [DOI 10.18653/v1/2026.acl-long.1080](https://doi.org/10.18653/v1/2026.acl-long.1080) · [OpenAlex](https://openalex.org/W7166848699)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
Large reasoning models tend to overthink: they keep re-verifying or trying other strategies after they already have a correct solution, which wastes tokens. Earlier early-exit methods (Dynasor-CoT, DEER, CGRS) probe intermediate answers and judge them by consistency or entropy. This is unreliable because models can be overconfident when wrong, and it only works when answers are short and have a fixed format. The authors propose DTSR (Dynamic Thought Sufficiency in Reasoning), a training-free decoding method modelled on human metacognition. During generation, the model watches for reflection cues such as "Wait". At each cue, it rates whether its chain-of-thought so far is enough to answer the question. If the score passes a threshold, the thinking phase is closed and the final answer is produced. Because DTSR needs no intermediate answer, it also applies to code and open-ended questions. On Qwen3 models, it cuts reasoning length by 28.9%–34.9% with minimal accuracy loss and beats the baselines. This makes inference cheaper without any retraining.

## Key points
- **Two stages.** (1) Reflection Signal Monitoring detects cue tokens such as "Wait", "Alternatively" and "But…" phrases. (2) Thought Sufficiency Check prompts the model to give a scalar sufficiency score for the current CoT. If the score exceeds threshold τ, `</think>` is appended; otherwise reasoning resumes.
- **Signal set origin.** The reflection signals come from an analysis of Qwen3-32B trajectories. Earliest-correct exit points were often followed by verification-style cues.
- **Check spacing.** A minimum token interval k between checks avoids repeated, redundant evaluations when cues appear close together.
- **Results.** Experiments used Qwen3 models of several sizes, up to 32B, on mathematics, programming and open-ended questions. Reasoning length dropped by 28.9%–34.9% with near-lossless performance.
- **Analysis and limits.** The paper also discusses LRM overconfidence and the self-evaluation paradigm, in which the model judges its own trace from a third-person view. Stated limitations are that no models above 32B were tested and that the work covers text-only tasks, with no multimodal or agent settings.

*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

- [Soochow University](https://md.research.elamur.ai/free/en/universities/soochow-university/) — University, Suzhou, China · 63 works · 4,507 citations
- [ByteDance](https://md.research.elamur.ai/free/en/companies/bytedance/) — Company, Beijing, China · 41 works · 4,180 citations

📚 Shelves (4): [🖥️ 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 · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

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

## Cite

Cite as: Yang Xiang, Yixin Ji, Ruotao Xu, Dan Qiao, Zheming Yang, Juntao Li and Min Zhang. 2026. When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

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

```bibtex
@inproceedings{xiang2026when,
  title = {{When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning}},
  author = {Yang Xiang, Yixin Ji, Ruotao Xu, Dan Qiao, Zheming Yang, Juntao Li and Min Zhang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1080},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1080}
}
```

- [← previous When Helpers Become Hazards: A Benchmark for Analyzing Multimodal LLM-Powered Safety in Daily Life](https://md.research.elamur.ai/free/en/papers/lou-2026-when-helpers-become-hazards-benchmark/)
- [next → When LLMs Annotate: Reliability Challenges in Low-Resource NLI](https://md.research.elamur.ai/free/en/papers/panahi-2026-when-llms-annotate-reliability-challenges/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/19/)

## References

49 references, 21 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Pranjal Aggarwal and Sean Welleck. 2025. [L1: Controlling how long a reasoning model thinks with reinforcement learning](https://md.research.elamur.ai/free/en/papers/aggarwal-2025-l1-controlling-long-reasoning-model/). Preprint, arXiv:2503.04697.
2. Daman Arora and Andrea Zanette. 2025a. Training language models to reason efficiently. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Daman+Arora+and+Andrea+Zanette%2E+2025a%2E+Training+language+models+to+reason+efficiently%2E+In+The+Thirty%2Dninth+Annual+Conference+on+Neural+Information+Processing+Systems%2E)
3. Daman Arora and Andrea Zanette. 2025b. Training language models to reason efficiently. CoRR, abs/2502.04463. [find](https://scholar.google.com/scholar?q=Daman+Arora+and+Andrea+Zanette%2E+2025b%2E+Training+language+models+to+reason+efficiently%2E+CoRR%2C+abs%2F2502%2E04463%2E)
4. Paul C. Bogdan, Uzay Macar, Neel Nanda, and Arthur Conmy. 2025. Thought anchors: Which llm reasoning steps matter? Preprint, arXiv:2506.19143. [find](https://scholar.google.com/scholar?q=Paul+C%2E+Bogdan%2C+Uzay+Macar%2C+Neel+Nanda%2C+and+Arthur+Conmy%2E+2025%2E+Thought+anchors%3A+Which+llm+reasoning+steps+matter%3F+Preprint%2C+arXiv%3A2506%2E19143%2E)
5. Qiguang Chen, Libo Qin, Jinhao Liu, Dengyun Peng, Jiannan Guan, Peng Wang, Mengkang Hu, Yuhang Zhou, Te Gao, and Wanxiang Che. 2025a. [Towards reasoning era: A survey of long chain-of-thought for reasoning large language models](https://md.research.elamur.ai/free/en/papers/chen-2025-towards-reasoning-era-survey-long/). CoRR, abs/2503.09567.
6. Xingyu Chen, Jiahao Xu, Tian Liang, Zhiwei He, Jianhui Pang, Dian Yu, Linfeng Song, Qiuzhi Liu, Mengfei Zhou, Zhuosheng Zhang, Rui Wang, Zhaopeng Tu, Haitao Mi, and Dong Yu. 2025b. Do NOT think that much for 2+3=? on the overthinking of long reasoning models. In Forty-second International Conference… [find](https://scholar.google.com/scholar?q=Xingyu+Chen%2C+Jiahao+Xu%2C+Tian+Liang%2C+Zhiwei+He%2C+Jianhui+Pang%2C+Dian+Yu%2C+Linfeng+Song%2C+Qiuzhi+Liu%2C+Mengfei+Zhou%2C+Zhuosheng+Zhang%2C+Rui+Wang%2C+Zhaopeng+Tu%2C+Haitao+Mi%2C+and+Dong+Yu%2E+2025b%2E+Do+NOT+think+that+m)
7. 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](https://md.research.elamur.ai/free/en/papers/cobbe-2021-training-verifiers-solve-math-word/). Preprint, arXiv:2110.14168.
8. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, and 1 others. 2025. [Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities](https://md.research.elamur.ai/free/en/papers/comanici-2025-gemini-2-5-pushing-frontier/). Pre…
9. Alejandro Cuadron, Dacheng Li, Wenjie Ma, Xingyao Wang, Yichuan Wang, Siyuan Zhuang, Shu Liu, Luis Gaspar Schroeder, Tian Xia, Huanzhi Mao, Nicholas Thumiger, Aditya Desai, Ion Stoica, Ana Klimovic, Graham Neubig, and Joseph E. Gonzalez. 2025. The danger of overthinking: Examining the reasoning-acti… [find](https://scholar.google.com/scholar?q=Alejandro+Cuadron%2C+Dacheng+Li%2C+Wenjie+Ma%2C+Xingyao+Wang%2C+Yichuan+Wang%2C+Siyuan+Zhuang%2C+Shu+Liu%2C+Luis+Gaspar+Schroeder%2C+Tian+Xia%2C+Huanzhi+Mao%2C+Nicholas+Thumiger%2C+Aditya+Desai%2C+Ion+Stoica%2C+Ana+Klimovic%2C+G)
10. Mz Dai, Chenxu Yang, and Qingyi Si. 2025. S-GRPO: Early exit via reinforcement learning in reasoning models. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Mz+Dai%2C+Chenxu+Yang%2C+and+Qingyi+Si%2E+2025%2E+S%2DGRPO%3A+Early+exit+via+reinforcement+learning+in+reasoning+models%2E+In+The+Thirty%2Dninth+Annual+Conference+on+Neural+Information+Processing+Systems%2E)
11. Gongfan Fang, Xinyin Ma, and Xinchao Wang. 2025. Thinkless: Llm learns when to think. Preprint, arXiv:2505.13379. — [arXiv:2505.13379](https://arxiv.org/abs/2505.13379) [find](https://scholar.google.com/scholar?q=Gongfan+Fang%2C+Xinyin+Ma%2C+and+Xinchao+Wang%2E+2025%2E+Thinkless%3A+Llm+learns+when+to+think%2E+Preprint%2C+arXiv%3A2505%2E13379%2E+%E2%80%94+%5BarXiv%3A2505%2E13379%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2505%2E13379%29)
12. Sicheng Feng, Gongfan Fang, Xinyin Ma, and Xinchao Wang. 2025. [Efficient reasoning models: A survey](https://md.research.elamur.ai/free/en/papers/feng-2025-efficient-reasoning-models-survey/). Preprint, arXiv:2504.10903.
13. John H. Flavell. 1979. Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist, 34:906–911. [find](https://scholar.google.com/scholar?q=John+H%2E+Flavell%2E+1979%2E+Metacognition+and+cognitive+monitoring%3A+A+new+area+of+cognitive%2Ddevelopmental+inquiry%2E+American+Psychologist%2C+34%3A906%E2%80%93911%2E)
14. D. S. Fleur, B. Bredeweg, and W. van den Bos. 2021. Metacognition: Ideas and insights from neuro- and educational sciences. npj Science of Learning, 6:1– 11. [find](https://scholar.google.com/scholar?q=D%2E+S%2E+Fleur%2C+B%2E+Bredeweg%2C+and+W%2E+van+den+Bos%2E+2021%2E+Metacognition%3A+Ideas+and+insights+from+neuro%2D+and+educational+sciences%2E+npj+Science+of+Learning%2C+6%3A1%E2%80%93+11%2E)
15. Yichao Fu, Junda Chen, Yonghao Zhuang, Zheyu Fu, Ion Stoica, and Hao Zhang. 2025. Reasoning without self-doubt: More efficient chain-of-thought through certainty probing. In ICLR 2025 Workshop on Foundation Models in the Wild. [find](https://scholar.google.com/scholar?q=Yichao+Fu%2C+Junda+Chen%2C+Yonghao+Zhuang%2C+Zheyu+Fu%2C+Ion+Stoica%2C+and+Hao+Zhang%2E+2025%2E+Reasoning+without+self%2Ddoubt%3A+More+efficient+chain%2Dof%2Dthought+through+certainty+probing%2E+In+ICLR+2025+Workshop+on+Foun)
16. Kanishk Gandhi, Ayush K Chakravarthy, Anikait Singh, Nathan Lile, and Noah Goodman. 2025. [Cognitive behaviors that enable self-improving reasoners, or, four habits of highly effective STars](https://md.research.elamur.ai/free/en/papers/gandhi-2025-cognitive-behaviors-that-enable-self/). In Second Conference on Language Modeling.
17. D. Guo, D. Yang, H. Zhang, and et al. 2025. Deepseekr1 incentivizes reasoning in llms through reinforcement learning. Nature, 645:633–638. [find](https://scholar.google.com/scholar?q=D%2E+Guo%2C+D%2E+Yang%2C+H%2E+Zhang%2C+and+et+al%2E+2025%2E+Deepseekr1+incentivizes+reasoning+in+llms+through+reinforcement+learning%2E+Nature%2C+645%3A633%E2%80%93638%2E)
18. Tingxu Han, Zhenting Wang, Chunrong Fang, Shiyu Zhao, Shiqing Ma, and Zhenyu Chen. 2025. [Token-budget-aware LLM reasoning](https://md.research.elamur.ai/free/en/papers/han-2024-token-budget-aware-llm-reasoning/). In Findings of the Association for Computational Linguistics: ACL 2025, pages 24842–24855, Vienna, Austria. Association for Computational Linguistics.
19. Chaoqun He, Renjie Luo, Yuzhuo Bai, Shengding Hu, Zhen Thai, Junhao Shen, Jinyi Hu, Xu Han, Yujie Huang, Yuxiang Zhang, Jie Liu, Lei Qi, Zhiyuan Liu, and Maosong Sun. 2024. [OlympiadBench: A challenging benchmark for promoting AGI with olympiad-level bilingual multimodal scientific problems](https://md.research.elamur.ai/free/en/papers/he-2024-olympiadbench-challenging-benchmark-promoting-agi/). In Proce…
20. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2021. [Measuring mathematical problem solving with the MATH dataset](https://md.research.elamur.ai/free/en/papers/hendrycks-2021-measuring-mathematical-problem-solving-math/). In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2).
21. Jiameng Huang, Baijiong Lin, Guhao Feng, Jierun Chen, Di He, and Lu Hou. 2025. Efficient reasoning for large reasoning language models via certainty-guided reflection suppression. Preprint, arXiv:2508.05337. — [arXiv:2508.05337](https://arxiv.org/abs/2508.05337) [find](https://scholar.google.com/scholar?q=Jiameng+Huang%2C+Baijiong+Lin%2C+Guhao+Feng%2C+Jierun+Chen%2C+Di+He%2C+and+Lu+Hou%2E+2025%2E+Efficient+reasoning+for+large+reasoning+language+models+via+certainty%2Dguided+reflection+suppression%2E+Preprint%2C+arXiv%3A2508)
22. 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, and 80 others. 20… — [OpenAI o1 System Card](https://md.research.elamur.ai/free/en/papers/openai-2024-openai-o1-system-card/)
23. Naman Jain, King Han, Alex Gu, Wen-Ding Li, Fanjia Yan, Tianjun Zhang, Sida Wang, Armando Solar-Lezama, Koushik Sen, and Ion Stoica. 2025. [Livecodebench: Holistic and contamination free evaluation of large language models for code](https://md.research.elamur.ai/free/en/papers/jain-2024-livecodebench-holistic-contamination-free-evaluation/). In The Thirteenth International Conference on Learning Representatio…
24. Yixin Ji, Juntao Li, Yang Xiang, Hai Ye, Kaixin Wu, Kai Yao, Jia Xu, Linjian Mo, and Min Zhang. 2025. A survey of test-time compute: From intuitive inference to deliberate reasoning. Preprint, arXiv:2501.02497. — [arXiv:2501.02497](https://arxiv.org/abs/2501.02497) [find](https://scholar.google.com/scholar?q=Yixin+Ji%2C+Juntao+Li%2C+Yang+Xiang%2C+Hai+Ye%2C+Kaixin+Wu%2C+Kai+Yao%2C+Jia+Xu%2C+Linjian+Mo%2C+and+Min+Zhang%2E+2025%2E+A+survey+of+test%2Dtime+compute%3A+From+intuitive+inference+to+deliberate+reasoning%2E+Preprint%2C+arXiv%3A2)
25. Guochao Jiang, Guofeng Quan, Zepeng Ding, Ziqin Luo, Dixuan Wang, and Zheng Hu. 2025a. Flash-think: An early exit method for efficient reasoning. Preprint, arXiv:2505.13949. — [arXiv:2505.13949](https://arxiv.org/abs/2505.13949) [find](https://scholar.google.com/scholar?q=Guochao+Jiang%2C+Guofeng+Quan%2C+Zepeng+Ding%2C+Ziqin+Luo%2C+Dixuan+Wang%2C+and+Zheng+Hu%2E+2025a%2E+Flash%2Dthink%3A+An+early+exit+method+for+efficient+reasoning%2E+Preprint%2C+arXiv%3A2505%2E13949%2E+%E2%80%94+%5BarXiv%3A2505%2E13949%5D%28https)
26. Lingjie Jiang, Xun Wu, Shaohan Huang, Qingxiu Dong, Zewen Chi, Li Dong, Xingxing Zhang, Tengchao Lv, Lei Cui, and Furu Wei. 2025b. Think only when you need with large hybrid-reasoning models. Preprint, arXiv:2505.14631. — [arXiv:2505.14631](https://arxiv.org/abs/2505.14631) [find](https://scholar.google.com/scholar?q=Lingjie+Jiang%2C+Xun+Wu%2C+Shaohan+Huang%2C+Qingxiu+Dong%2C+Zewen+Chi%2C+Li+Dong%2C+Xingxing+Zhang%2C+Tengchao+Lv%2C+Lei+Cui%2C+and+Furu+Wei%2E+2025b%2E+Think+only+when+you+need+with+large+hybrid%2Dreasoning+models%2E+Preprint)
27. Ayeong Lee, Ethan Che, and Tianyi Peng. 2025. How well do LLMs compress their own chain-of-thought? a token complexity approach. In ES-FoMo III: 3rd Workshop on Efficient Systems for Foundation Models. [find](https://scholar.google.com/scholar?q=Ayeong+Lee%2C+Ethan+Che%2C+and+Tianyi+Peng%2E+2025%2E+How+well+do+LLMs+compress+their+own+chain%2Dof%2Dthought%3F+a+token+complexity+approach%2E+In+ES%2DFoMo+III%3A+3rd+Workshop+on+Efficient+Systems+for+Foundation+Models)
28. Tengxiao Liu, Qipeng Guo, Xiangkun Hu, Cheng Jiayang, Yue Zhang, Xipeng Qiu, and Zheng Zhang. 2024. Can language models learn to skip steps? In The Thirty-eighth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Tengxiao+Liu%2C+Qipeng+Guo%2C+Xiangkun+Hu%2C+Cheng+Jiayang%2C+Yue+Zhang%2C+Xipeng+Qiu%2C+and+Zheng+Zhang%2E+2024%2E+Can+language+models+learn+to+skip+steps%3F+In+The+Thirty%2Deighth+Annual+Conference+on+Neural+Informatio)
29. Xin Liu and Lu Wang. 2025. [Answer convergence as a signal for early stopping in reasoning](https://md.research.elamur.ai/free/en/papers/liu-2025-answer-convergence-as-signal-early/). Preprint, arXiv:2506.02536.
30. Chenwei Lou, Zewei Sun, Xinnian Liang, Meng Qu, Wei Shen, Wenqi Wang, Yuntao Li, Qingping Yang, and Shuangzhi Wu. 2025. Adacot: Pareto-optimal adaptive chain-of-thought triggering via reinforcement learning. Preprint, arXiv:2505.11896. — [arXiv:2505.11896](https://arxiv.org/abs/2505.11896) [find](https://scholar.google.com/scholar?q=Chenwei+Lou%2C+Zewei+Sun%2C+Xinnian+Liang%2C+Meng+Qu%2C+Wei+Shen%2C+Wenqi+Wang%2C+Yuntao+Li%2C+Qingping+Yang%2C+and+Shuangzhi+Wu%2E+2025%2E+Adacot%3A+Pareto%2Doptimal+adaptive+chain%2Dof%2Dthought+triggering+via+reinforcement+le)
31. Wenjie Ma, Jingxuan He, Charlie Snell, Tyler Griggs, Sewon Min, and Matei Zaharia. 2025a. Reasoning models can be effective without thinking. Preprint, arXiv:2504.09858. [find](https://scholar.google.com/scholar?q=Wenjie+Ma%2C+Jingxuan+He%2C+Charlie+Snell%2C+Tyler+Griggs%2C+Sewon+Min%2C+and+Matei+Zaharia%2E+2025a%2E+Reasoning+models+can+be+effective+without+thinking%2E+Preprint%2C+arXiv%3A2504%2E09858%2E)
32. Xinyin Ma, Guangnian Wan, Runpeng Yu, Gongfan Fang, and Xinchao Wang. 2025b. [CoT-valve: Length-compressible chain-of-thought tuning](https://md.research.elamur.ai/free/en/papers/ma-2025-cot-valve-length-compressible-chain/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6025–6035, Vienna, Austria. Association for C…
33. Zhiting Mei, Christina Zhang, Tenny Yin, Justin Lidard, Ola Shorinwa, and Anirudha Majumdar. 2025. [Reasoning about uncertainty: Do reasoning models know when they don’t know?](https://md.research.elamur.ai/free/en/papers/mei-2026-reasoning-about-uncertainty-do-reasoning/) Preprint, arXiv:2506.18183.
34. Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candes, and Tatsunori Hashimoto. 2025. [s1: Simple test-time scaling](https://md.research.elamur.ai/free/en/papers/muennighoff-2025-s1-simple-test-time-scaling/). In Workshop on Reasoning and Planning for Large Language Models.
35. OpenAI. 2024. [Openai o1 system card](https://md.research.elamur.ai/free/en/papers/openai-2024-openai-o1-system-card/). Preprint, arXiv:2412.16720.
36. OpenAI. 2025. Introducing openai o3 and o4-mini. [find](https://scholar.google.com/scholar?q=OpenAI%2E+2025%2E+Introducing+openai+o3+and+o4%2Dmini%2E)
37. Ziqing Qiao, Yongheng Deng, Jiali Zeng, Dong Wang, Lai Wei, Fandong Meng, Jie Zhou, Ju Ren, and Yaoxue Zhang. 2025. [Concise: Confidence-guided compression in step-by-step efficient reasoning](https://md.research.elamur.ai/free/en/papers/qiao-2025-concise-confidence-guided-compression-step/). Preprint, arXiv:2505.04881.
38. David Rein, Betty Li Hou, Asa Cooper Stickland, Jackson Petty, Richard Yuanzhe Pang, Julien Dirani, Julian Michael, and Samuel R. Bowman. 2024. [GPQA: A graduate-level google-proof q&a benchmark](https://md.research.elamur.ai/free/en/papers/rein-2023-gpqa-graduate-level-google-proof/). In First Conference on Language Modeling.
39. Matthew Renze and Erhan Guven. 2024. The benefits of a concise chain of thought on problem-solving in large language models. In 2024 2nd International Conference on Foundation and Large Language Models (FLLM), page 476–483. IEEE. [find](https://scholar.google.com/scholar?q=Matthew+Renze+and+Erhan+Guven%2E+2024%2E+The+benefits+of+a+concise+chain+of+thought+on+problem%2Dsolving+in+large+language+models%2E+In+2024+2nd+International+Conference+on+Foundation+and+Large+Language+Model)
40. Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, Chuning Tang, Congcong Wang, Dehao Zhang, Enming Yuan, Enzhe Lu, Fengxiang Tang, Flood Sung, Guangda Wei, Guokun Lai, and 75 others. 2025. [Kimi k1.5: Scaling reinforcement l…](https://md.research.elamur.ai/free/en/papers/kimi-team-2025-kimi-k1-5-scaling-reinforcement/)
41. Chenlong Wang, Yuanning Feng, Dongping Chen, Zhaoyang Chu, Ranjay Krishna, and Tianyi Zhou. 2025. Wait, we don’t need to "wait"! removing thinking tokens improves reasoning efficiency. Preprint, arXiv:2506.08343. [find](https://scholar.google.com/scholar?q=Chenlong+Wang%2C+Yuanning+Feng%2C+Dongping+Chen%2C+Zhaoyang+Chu%2C+Ranjay+Krishna%2C+and+Tianyi+Zhou%2E+2025%2E+Wait%2C+we+don%E2%80%99t+need+to+%22wait%22%21+removing+thinking+tokens+improves+reasoning+efficiency%2E+Preprint%2C+arXiv)
42. Silei Xu, Wenhao Xie, Lingxiao Zhao, and Pengcheng He. 2025a. Chain of draft: Thinking faster by writing less. Preprint, arXiv:2502.18600. [find](https://scholar.google.com/scholar?q=Silei+Xu%2C+Wenhao+Xie%2C+Lingxiao+Zhao%2C+and+Pengcheng+He%2E+2025a%2E+Chain+of+draft%3A+Thinking+faster+by+writing+less%2E+Preprint%2C+arXiv%3A2502%2E18600%2E)
43. Yuhui Xu, Hanze Dong, Lei Wang, Doyen Sahoo, Junnan Li, and Caiming Xiong. 2025b. Scalable chain of thoughts via elastic reasoning. Preprint, arXiv:2505.05315. — [arXiv:2505.05315](https://arxiv.org/abs/2505.05315) [find](https://scholar.google.com/scholar?q=Yuhui+Xu%2C+Hanze+Dong%2C+Lei+Wang%2C+Doyen+Sahoo%2C+Junnan+Li%2C+and+Caiming+Xiong%2E+2025b%2E+Scalable+chain+of+thoughts+via+elastic+reasoning%2E+Preprint%2C+arXiv%3A2505%2E05315%2E+%E2%80%94+%5BarXiv%3A2505%2E05315%5D%28https%3A%2F%2Farxiv%2Eorg%2Fa)
44. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 41 others. 2025a. [Qwen3 technical report](https://md.research.elamur.ai/free/en/papers/yang-2025-qwen3-technical-report/). Preprint, arXiv:2505.09388.
45. Chenxu Yang, Qingyi Si, Yongjie Duan, Zheliang Zhu, Chenyu Zhu, Qiaowei Li, Zheng Lin, Li Cao, and Weiping Wang. 2025b. Dynamic early exit in reasoning models. Preprint, arXiv:2504.15895. [find](https://scholar.google.com/scholar?q=Chenxu+Yang%2C+Qingyi+Si%2C+Yongjie+Duan%2C+Zheliang+Zhu%2C+Chenyu+Zhu%2C+Qiaowei+Li%2C+Zheng+Lin%2C+Li+Cao%2C+and+Weiping+Wang%2E+2025b%2E+Dynamic+early+exit+in+reasoning+models%2E+Preprint%2C+arXiv%3A2504%2E15895%2E)
46. Edward Yeo, Yuxuan Tong, Xinyao Niu, Graham Neubig, and Xiang Yue. 2025. [Demystifying long chain-of-thought reasoning in LLMs](https://md.research.elamur.ai/free/en/papers/yeo-2025-demystifying-long-chain-thought-reasoning/). In ICLR 2025 Workshop on Deep Generative Model in Machine Learning: Theory, Principle and Efficacy.
47. Jingyang Yi, Jiazheng Wang, and Sida Li. 2025. Short-erbetter: Guiding reasoning models to find optimal inference length for efficient reasoning. Preprint, arXiv:2504.21370. — [arXiv:2504.21370](https://arxiv.org/abs/2504.21370) [find](https://scholar.google.com/scholar?q=Jingyang+Yi%2C+Jiazheng+Wang%2C+and+Sida+Li%2E+2025%2E+Short%2Derbetter%3A+Guiding+reasoning+models+to+find+optimal+inference+length+for+efficient+reasoning%2E+Preprint%2C+arXiv%3A2504%2E21370%2E+%E2%80%94+%5BarXiv%3A2504%2E21370%5D%28https)
48. Ping Yu, Jing Xu, Jason E Weston, and Ilia Kulikov. 2024. Distilling system 2 into system 1. In The First Workshop on System-2 Reasoning at Scale, NeurIPS’24. [find](https://scholar.google.com/scholar?q=Ping+Yu%2C+Jing+Xu%2C+Jason+E+Weston%2C+and+Ilia+Kulikov%2E+2024%2E+Distilling+system+2+into+system+1%2E+In+The+First+Workshop+on+System%2D2+Reasoning+at+Scale%2C+NeurIPS%E2%80%9924%2E)
49. Anqi Zhang, Yulin Chen, Jane Pan, Chen Zhao, Aurojit Panda, Jinyang Li, and He He. 2025. Reasoning models know when they’re right: Probing hidden states for self-verification. In Second Conference on Language Modeling. [find](https://scholar.google.com/scholar?q=Anqi+Zhang%2C+Yulin+Chen%2C+Jane+Pan%2C+Chen+Zhao%2C+Aurojit+Panda%2C+Jinyang+Li%2C+and+He+He%2E+2025%2E+Reasoning+models+know+when+they%E2%80%99re+right%3A+Probing+hidden+states+for+self%2Dverification%2E+In+Second+Conference+on+)

*Updated 31 October 2026*