Elamur Research⌕
☰
home / papers / 2025 / Reasoning Models Can Be Effective Without Thinking

Reasoning Models Can Be Effective Without Thinking

Paper · 2025 · Wenjie Ma, Jingxuan He, Charlie Snell et al.

Summary

Large reasoning models usually write a long chain of thought before answering, and this costs many tokens and adds latency. The authors ask whether that step is actually needed. Using DeepSeek-R1-Distill-Qwen, they introduce NoThinking. The assistant turn is prefilled with a short placeholder thinking block, so the model goes straight to the final solution. The method needs no extra training.

They compare NoThinking with standard Thinking on seven benchmarks covering competition math, coding and formal theorem proving. Budget forcing equalizes token usage, and pass@k is the main metric. Without thinking, the model uses far fewer tokens and wins clearly at small budgets. Any gap it shows at low k shrinks or reverses as k grows.

The authors exploit this by sampling several NoThinking answers in parallel and picking one with a verifier or a confidence-based best-of-N rule. This matches or beats sequential Thinking at much lower latency. The work challenges the assumption that long reasoning traces are essential and offers a cheap, strong baseline for low-budget or latency-sensitive settings.

Key points

  • Benchmarks: AIME 2024, AIME 2025, AMC 2023, OlympiadBench, LiveCodeBench, MiniF2F and ProofNet, evaluated with pass@k for k from 1 to 64.
  • Token savings and low-budget wins: Without budget control, NoThinking needs 2.0–5.1× fewer tokens than Thinking. With a 700-token budget on AMC 2023, it scores 51.3 against 28.9 for Thinking.
  • Tasks with verifiers: On the theorem-proving tasks (MiniF2F, ProofNet), parallel NoThinking reaches Thinking-level accuracy with about 7× lower latency and roughly 4× fewer output tokens.
  • Tasks without verifiers: On OlympiadBench, parallel NoThinking with confidence-based selection scores 55.79 versus 54.1 for unrestricted Thinking, with about 9× lower latency. Confidence-based selection generally beats plain majority voting.
  • Weak spot: LiveCodeBench is the exception. Coding outputs cannot be grouped by exact match for voting, so selection falls back to the single most confident answer, which is less reliable.

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 California, BerkeleyBerkeley · United States243 works · 251,774 citations🔬Allen Institute for Artificial IntelligenceSeattle · United States317 works · 106,633 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksArts and Humanities · 17,959 worksLanguage and Linguistics · 17,789 worksComputer Vision and Pattern Recognition · 3,621 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,948 works

Cite

Cite as Wenjie Ma, Jingxuan He, Charlie Snell, Tyler Griggs, Sewon Min and Matei Zaharia. 2025. Reasoning Models Can Be Effective Without Thinking. arXiv:2504.09858.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2504.09858 · arXiv 2504.09858

@misc{ma2025reasoning,
  title = {{Reasoning Models Can Be Effective Without Thinking}},
  author = {Wenjie Ma, Jingxuan He, Charlie Snell, Tyler Griggs, Sewon Min and Matei Zaharia},
  year = {2025},
  doi = {10.48550/arxiv.2504.09858},
  eprint = {2504.09858},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2504.09858}
}

References

59 references, 37 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. L1: Controlling how long a reasoning model thinks with reinforcement learning, 2025. URL https://arxiv.org/abs/2503.04697. ↗
  2. Daman Arora and Andrea Zanette. Training language models to reason efficiently, 2025. URL https://arxiv.org/abs/2502.04463. ↗
  3. Simon A. Aytes, Jinheon Baek, and Sung Ju Hwang. Sketch-of-thought: Efficient llm reasoning with adaptive cognitive-inspired sketching, 2025. URL https://arxiv.org/ abs/2503.05179. ↗
  4. Zhangir Azerbayev, Bartosz Piotrowski, Hailey Schoelkopf, Edward W. Ayers, Dragomir Radev, and Jeremy Avigad. Proofnet: Autoformalizing and formally proving undergraduate-level mathematics, 2023. URL https://arxiv.org/abs/2302.12433. ↗
  5. Bradley Brown, Jordan Juravsky, Ryan Ehrlich, Ronald Clark, Quoc V. Le, Christopher Ré, and Azalia Mirhoseini. Large language monkeys: Scaling inference compute with repeated sampling, 2024. URL https://arxiv.org/abs/2407.21787. ↗
  6. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, Nick Ryder, Mikhai… — Evaluating Large Language Models Trained on Code
  7. Jeffrey Cheng and Benjamin Van Durme. Compressed chain of thought: Efficient reasoning through dense representations, 2024. URL https://arxiv.org/abs/2412.13171. ↗
  8. Yu-Neng Chuang, Leisheng Yu, Guanchu Wang, Lizhe Zhang, Zirui Liu, Xuanting Cai, Yang Sui, Vladimir Braverman, and Xia Hu. Confident or seek stronger: Exploring uncertainty-based on-device llm routing from benchmarking to generalization, 2025a. URL https://arxiv.org/abs/2502.04428. — [arXiv:2502.04428](https://arxiv.org/abs/2502.04428) ↗
  9. Yu-Neng Chuang, Helen Zhou, Prathusha Kameswara Sarma, Parikshit Gopalan, John Boccio, Sara Bolouki, and Xia Hu. Learning to route llms with confidence tokens, 2025b. URL https://arxiv.org/abs/2410.13284. — [arXiv:2410.13284](https://arxiv.org/abs/2410.13284) ↗
  10. Yingqian Cui, Pengfei He, Jingying Zeng, Hui Liu, Xianfeng Tang, Zhenwei Dai, Yan Han, Chen Luo, Jing Huang, Zhen Li, Suhang Wang, Yue Xing, Jiliang Tang, and Qi He. Stepwise perplexity-guided refinement for efficient chain-of-thought reasoning in large language models, 2025. URL https://arxiv.org/a… ↗
  11. 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… — DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
  12. Yifu Ding, Wentao Jiang, Shunyu Liu, Yongcheng Jing, Jinyang Guo, Yingjie Wang, Jing Zhang, Zengmao Wang, Ziwei Liu, Bo Du, Xianglong Liu, and Dacheng Tao. Dynamic parallel tree search for efficient llm reasoning, 2025. URL https://arxiv.org/abs/2502. 16235. ↗
  13. Yichao Fu, Junda Chen, Siqi Zhu, Zheyu Fu, Zhongdongming Dai, Aurick Qiao, and Hao Zhang. Efficiently serving llm reasoning programs with certaindex, 2024. URL https://arxiv.org/abs/2412.20993. — [arXiv:2412.20993](https://arxiv.org/abs/2412.20993) ↗
  14. Tingxu Han, Zhenting Wang, Chunrong Fang, Shiyu Zhao, Shiqing Ma, and Zhenyu Chen. Token-budget-aware llm reasoning, 2025. URL https://arxiv.org/abs/2412.18547. ↗
  15. Shibo Hao, Sainbayar Sukhbaatar, DiJia Su, Xian Li, Zhiting Hu, Jason Weston, and Yuandong Tian. Training large language models to reason in a continuous latent space, 2024. URL https://arxiv.org/abs/2412.06769. ↗
  16. Chaoqun He, Renjie Luo, Yuzhuo Bai, Shengding Hu, Zhen Leng Thai, Junhao Shen, Jinyi Hu, Xu Han, Yujie Huang, Yuxiang Zhang, Jie Liu, Lei Qi, Zhiyuan Liu, and Maosong Sun. Olympiadbench: A challenging benchmark for promoting agi with olympiad-level bilingual multimodal scientific problems, 2024. URL…
  17. Naman Jain, King Han, Alex Gu, Wen-Ding Li, Fanjia Yan, Tianjun Zhang, Sida Wang, Armando Solar-Lezama, Koushik Sen, and Ion Stoica. Livecodebench: Holistic and contamination free evaluation of large language models for code. In The Thirteenth International Conference on Learning Representations, 20…
  18. Yu Kang, Xianghui Sun, Liangyu Chen, and Wei Zou. C3ot: Generating shorter chain-of-thought without compromising effectiveness, 2024. URL https://arxiv.org/abs/2412. 11664. ↗
  19. Zhewei Kang, Xuandong Zhao, and Dawn Song. Scalable best-of-n selection for large language models via self-certainty, 2025. URL https://arxiv.org/abs/2502.18581. ↗
  20. Nathan Lambert, Jacob Morrison, Valentina Pyatkin, Shengyi Huang, Hamish Ivison, Faeze Brahman, Lester James V. Miranda, Alisa Liu, Nouha Dziri, Shane Lyu, Yuling Gu, Saumya Malik, Victoria Graf, Jena D. Hwang, Jiangjiang Yang, Ronan Le Bras, Oyvind Tafjord, Chris Wilhelm, Luca Soldaini, Noah A. Smi… — Tulu 3: Pushing Frontiers in Open Language Model Post-Training
  21. Ayeong Lee, Ethan Che, and Tianyi Peng. How well do llms compress their own chain-of-thought? a token complexity approach, 2025. URL https://arxiv.org/abs/2503.01141. — [arXiv:2503.01141](https://arxiv.org/abs/2503.01141) ↗
  22. Dacheng Li, Shiyi Cao, Tyler Griggs, Shu Liu, Xiangxi Mo, Eric Tang, Sumanth Hegde, Kourosh Hakhamaneshi, Shishir G. Patil, Matei Zaharia, Joseph E. Gonzalez, and Ion Stoica. Llms can easily learn to reason from demonstrations structure, not content, is what matters!, 2025a. URL https://arxiv.org/ab… ↗
  23. Peiji Li, Kai Lv, Yunfan Shao, Yichuan Ma, Linyang Li, Xiaoqing Zheng, Xipeng Qiu, and Qipeng Guo. Fastmcts: A simple sampling strategy for data synthesis, 2025b. URL https://arxiv.org/abs/2502.11476. ↗
  24. Baohao Liao, Yuhui Xu, Hanze Dong, Junnan Li, Christof Monz, Silvio Savarese, Doyen Sahoo, and Caiming Xiong. Reward-guided speculative decoding for efficient llm reasoning, 2025. URL https://arxiv.org/abs/2501.19324. — [arXiv:2501.19324](https://arxiv.org/abs/2501.19324) ↗
  25. Tengxiao Liu, Qipeng Guo, Xiangkun Hu, Cheng Jiayang, Yue Zhang, Xipeng Qiu, and Zheng Zhang. Can language models learn to skip steps?, 2024. URL https://arxiv.org/ abs/2411.01855. — [arXiv:2411.01855](https://arxiv.org/abs/2411.01855) ↗
  26. Haotian Luo, Li Shen, Haiying He, Yibo Wang, Shiwei Liu, Wei Li, Naiqiang Tan, Xiaochun Cao, and Dacheng Tao. O1-pruner: Length-harmonizing fine-tuning for o1-like reasoning pruning, 2025. URL https://arxiv.org/abs/2501.12570. ↗
  27. Xinyin Ma, Guangnian Wan, Runpeng Yu, Gongfan Fang, and Xinchao Wang. Cot-valve: Length-compressible chain-of-thought tuning, 2025. URL https://arxiv.org/abs/2502. 09601. ↗
  28. Leonardo de Moura and Sebastian Ullrich. The lean 4 theorem prover and programming language. In Automated Deduction – CADE 28: 28th International Conference on Automated Deduction, Virtual Event, July 12–15, 2021, Proceedings, pp. 625–635, Berlin, Heidelberg, 2021. Springer-Verlag. ISBN 978-3-030-79…
  29. Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candès, and Tatsunori Hashimoto. s1: Simple test-time scaling, 2025. URL https://arxiv.org/abs/2501.19393. ↗
  30. Tergel Munkhbat, Namgyu Ho, Seo Hyun Kim, Yongjin Yang, Yujin Kim, and Se-Young Yun. Self-training elicits concise reasoning in large language models, 2025. URL https://arxiv.org/abs/2502.20122. ↗
  31. Isaac Ong, Amjad Almahairi, Vincent Wu, Wei-Lin Chiang, Tianhao Wu, Joseph E. Gonzalez, M Waleed Kadous, and Ion Stoica. Routellm: Learning to route llms with preference data, 2025. URL https://arxiv.org/abs/2406.18665. ↗
  32. OpenAI. Learning to reason with llms. OpenAI, 2024. URL https://openai.com/index/ learning-to-reason-with-llms/. ↗
  33. Yuxiao Qu, Matthew Y. R. Yang, Amrith Setlur, Lewis Tunstall, Edward Emanuel Beeching, Ruslan Salakhutdinov, and Aviral Kumar. Optimizing test-time compute via meta reinforcement fine-tuning, 2025. URL https://arxiv.org/abs/2503.07572. ↗
  34. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng… find
  35. Qwen Team. Preview of qwen qwen1.5-32b. qwq-32b-preview/, 2025. Accessed: 2025-03-20. https://qwenlm.github.io/blog/ ↗
  36. Matthew Renze and Erhan Guven. 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), pp. 476–483. find
  37. IEEE, November 2024. doi: 10.1109/fllm63129. 2024.10852493. URL http://dx.doi.org/10.1109/FLLM63129.2024.10852493. — [doi:10.1109/fllm63129](https://doi.org/10.1109/fllm63129) ↗
  38. Nikunj Saunshi, Nishanth Dikkala, Zhiyuan Li, Sanjiv Kumar, and Sashank J. Reddi. Reasoning with latent thoughts: On the power of looped transformers, 2025. URL https://arxiv.org/abs/2502.17416. ↗
  39. Xuan Shen, Yizhou Wang, Xiangxi Shi, Yanzhi Wang, Pu Zhao, and Jiuxiang Gu. Efficient reasoning with hidden thinking, 2025a. URL https://arxiv.org/abs/2501.19201. — [arXiv:2501.19201](https://arxiv.org/abs/2501.19201) ↗
  40. Yi Shen, Jian Zhang, Jieyun Huang, Shuming Shi, Wenjing Zhang, Jiangze Yan, Ning Wang, Kai Wang, and Shiguo Lian. Dast: Difficulty-adaptive slow-thinking for large reasoning models, 2025b. URL https://arxiv.org/abs/2503.04472. ↗
  41. Zhenyi Shen, Hanqi Yan, Linhai Zhang, Zhanghao Hu, Yali Du, and Yulan He. Codi: Compressing chain-of-thought into continuous space via self-distillation, 2025c. URL https://arxiv.org/abs/2502.21074. ↗
  42. Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. Scaling llm test-time compute optimally can be more effective than scaling model parameters, 2024. URL https://arxiv.org/abs/2408.03314. ↗
  43. DiJia Su, Hanlin Zhu, Yingchen Xu, Jiantao Jiao, Yuandong Tian, and Qinqing Zheng. Token assorted: Mixing latent and text tokens for improved language model reasoning, 2025. URL https://arxiv.org/abs/2502.03275. — [arXiv:2502.03275](https://arxiv.org/abs/2502.03275) ↗
  44. Yang Sui, Yu-Neng Chuang, Guanchu Wang, Jiamu Zhang, Tianyi Zhang, Jiayi Yuan, Hongyi Liu, Andrew Wen, Shaochen Zhong, Hanjie Chen, and Xia Hu. Stop overthinking: A survey on efficient reasoning for large language models, 2025. URL https://arxiv.org/ abs/2503.16419. ↗
  45. Hanshi Sun, Momin Haider, Ruiqi Zhang, Huitao Yang, Jiahao Qiu, Ming Yin, Mengdi Wang, Peter Bartlett, and Andrea Zanette. Fast best-of-n decoding via speculative rejection, 2024. URL https://arxiv.org/abs/2410.20290. ↗
  46. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models, 2023. URL https://arxiv.org/abs/2203.11171. ↗
  47. Yiming Wang, Pei Zhang, Siyuan Huang, Baosong Yang, Zhuosheng Zhang, Fei Huang, and Rui Wang. Sampling-efficient test-time scaling: Self-estimating the best-of-n sampling in early decoding, 2025. URL https://arxiv.org/abs/2503.01422. ↗
  48. Tong Wu, Chong Xiang, Jiachen T. Wang, and Prateek Mittal. Effectively controlling reasoning models through thinking intervention, 2025a. URL https://arxiv.org/abs/ 2503.24370. ↗
  49. Yuyang Wu, Yifei Wang, Tianqi Du, Stefanie Jegelka, and Yisen Wang. When more is less: Understanding chain-of-thought length in llms, 2025b. URL https://arxiv.org/abs/ 2502.07266. ↗
  50. Heming Xia, Yongqi Li, Chak Tou Leong, Wenjie Wang, and Wenjie Li. Tokenskip: Controllable chain-of-thought compression in llms, 2025. URL https://arxiv.org/abs/2502. 12067. ↗
  51. Silei Xu, Wenhao Xie, Lingxiao Zhao, and Pengcheng He. Chain of draft: Thinking faster by writing less, 2025a. URL https://arxiv.org/abs/2502.18600. ↗
  52. Yige Xu, Xu Guo, Zhiwei Zeng, and Chunyan Miao. Softcot: Soft chain-of-thought for efficient reasoning with llms, 2025b. URL https://arxiv.org/abs/2502.12134. ↗
  53. Yuchen Yan, Yongliang Shen, Yang Liu, Jin Jiang, Mengdi Zhang, Jian Shao, and Yueting Zhuang. Inftythink: Breaking the length limits of long-context reasoning in large language models, 2025. URL https://arxiv.org/abs/2503.06692. ↗
  54. Yixin Ye, Zhen Huang, Yang Xiao, Ethan Chern, Shijie Xia, and Pengfei Liu. Limo: Less is more for reasoning, 2025. URL https://arxiv.org/abs/2502.03387. ↗
  55. Ping Yu, Jing Xu, Jason Weston, and Ilia Kulikov. Distilling system 2 into system 1, 2024. URL https://arxiv.org/abs/2407.06023. ↗
  56. Lifan Yuan, Wendi Li, Huayu Chen, Ganqu Cui, Ning Ding, Kaiyan Zhang, Bowen Zhou, Zhiyuan Liu, and Hao Peng. Free process rewards without process labels, 2024. URL https://arxiv.org/abs/2412.01981. ↗
  57. Eric Zelikman, Yuhuai Wu, Jesse Mu, and Noah D. Goodman. Star: Bootstrapping reasoning with reasoning, 2022. URL https://arxiv.org/abs/2203.14465. ↗
  58. Jintian Zhang, Yuqi Zhu, Mengshu Sun, Yujie Luo, Shuofei Qiao, Lun Du, Da Zheng, Huajun Chen, and Ningyu Zhang. Lightthinker: Thinking step-by-step compression, 2025. URL https://arxiv.org/abs/2502.15589. ↗
  59. Kunhao Zheng, Jesse Michael Han, and Stanislas Polu. minif2f: a cross-system benchmark for formal olympiad-level mathematics. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=9ZPegFuFTFv. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ma-2025-reasoning-models-effective-thinking/