Early Stopping in Large Reasoning Models: A Survey and Empirical Audit of Efficiency and Reliability

Quang Hung Nguyen; Hanh Nguyen; Thanh Tam Nguyen; Quoc Viet Hung Nguyen; Trinh Pham; Thanh Trung Huynh

cs.AIOct 30, 2026

Early Stopping in Large Reasoning Models: A Survey and Empirical Audit of Efficiency and Reliability

Authors: Quang Hung Nguyen, Hanh Nguyen, Thanh Tam Nguyen, Quoc Viet Hung Nguyen, Trinh Pham, Thanh Trung Huynh

Organizations: VinUniversity, Vietnam · Griffith University, Australia · VinUni Bigdata Research Institute, VinUniversity, Vietnam

Abstract

Large reasoning models can improve accuracy by generating longer reasoning traces, but some of this computation may be unnecessary. Early-stopping methods aim to stop once enough reasoning is available to answer. Evaluating these methods requires accounting for the cost of deciding when to stop and the reliability of the resulting answer. We evaluate different methods across five reasoning models on mathematical and scientific reasoning, code generation, and factuality tasks by using a shared rollout replay protocol. We introduce Target-Integrated Token Efficiency (TITE) to measure efficiency relative to the oracle across accuracy targets. We also examine correctness at later exit positions, transfer across models and tasks, post-stop reasoning continuation, and retained justification quality. In our experiments, correctness can decrease at later positions beyond what the fitted decoding-noise model predicts. Reliability checks also reveal continued reasoning after forced termination and weaker supporting reasoning in some correct early-stopping outputs. The implementation can be found in https://github.com/hung20gg/audit-early-stopping.

Figures & tables

	Task Coverage					Evaluation Dimensions
Study	Math	Sci. QA	Gen. QA	Complex Code	Fact.	Oracle	Exit Dynamics	Post-stop Audit	CoT Quality
Han et al. (2025)	$\checkmark$					$\triangle$	$\triangle$
Wang et al. (2025b)	$\checkmark$					$\triangle$			$\checkmark$
Wu et al. (2025)	$\checkmark$	$\checkmark$
Dai et al. (2025)	$\checkmark$	$\checkmark$					$\triangle$
Liu and Wang (2025)	$\checkmark$	$\checkmark$	$\checkmark$				$\checkmark$		$\checkmark$

Table 1: Comparison with closely related work. Checkmarks denote evaluation components explicitly supported by the study.

\triangle

denotes related evidence that does not meet the exact criterion.

Panel of Figure 1 — Figure 1: Overview of the replay-based benchmark . (I) The frozen reasoning model answers each question once, and the full trajectory is stored. (II) The stored trajectory is replayed. At each checkpoint, the prefix followed by </think> yields one shared probe, and state-based, answer-free and answer-based verifiers read either the hidden states or this probe. Each method’s stopping rule then decides to stop or to continue to the next checkpoint. (III) A method that stops generates its final answer, which is checked against the gold answer and compared with reference policies in accuracy and total token cost.

Figure 2: Random exit accuracy for MATH500 problem 272 using gpt-oss-20b. Token colors indicate accuracy, ranging from red (0) to green (1) .

Panel of Figure 3 — Figure 3: Scoring for LiveCodeBench question 585: minimum-cost-for-cutting-cake-i (substring only). The two $\blacksquare$ markers indicate token limits of 20 and 100, respectively. Green highlights tokens that exceed the corresponding thresholds. The first 20 tokens of both solutions consist of imports and function declarations, while the scores only fluctuate in later sections

Dataset	Samples	Format
AIME26	30	Free-text
MATH500	500	Free-text
GSM8K	1,319	Free-text
GPQA-Diamond	198	MCQ
MMLU-Pro (subset)	2,406	MCQ
LiveCodeBench v6 Lite (wo tools)	1,055	Test cases

Table 2: Summary of evaluation datasets.

Figure 4: Example candidate exit positions along a reasoning trajectory.

Method	Gemma4 E4B		gpt-oss-20b		Qwen3.5-4B		Qwen3.5-9B		R1-Qwen3-8B
	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE
Optimal bucket	0.701	–	0.894	–	0.865	–	0.891	–	0.858	–
Full Thinking	0.590	–	0.824	–	0.802	–	0.821	–	0.749	–
No thinking	0.297	–	0.381	–	0.469	–	0.513	–	0.297	–
Statistical EE	0.590	0.403	0.824	0.191	0.802	0.151	0.841	0.176	0.749	0.177
EntroCut	0.574	0.404	0.773	0.380	0.700	0.269	0.741	0.207	0.709	0.354

Table 3: Average accuracy and TITE on the Mathematical & Science benchmarks.

Method	Gemma4 E4B		gpt-oss-20b		Qwen3.5-4B		Qwen3.5-9B		R1-Qwen3-8B
	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE
Optimal bucket	0.443	–	0.624	–	0.518	–	0.624	–	0.514	–
Full Thinking	0.415	–	0.510	–	0.424	–	0.530	–	0.382	–
No thinking	0.278	–	0.385	–	0.254	–	0.356	–	0.191	–
Statistical EE	0.416	0.303	0.514	0.038	0.426	0.052	0.535	0.081	0.379	0.048
EntroCut	0.397	0.297	0.478	0.098	0.323	0.046	0.385	0.015	0.267	0.025

Table 4: Average accuracy and TITE on the LiveCodeBench.

Method	Gemma4 E4B		gpt-oss-20b		Qwen3.5-4B		Qwen3.5-9B		R1-Qwen3-8B
	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE	Acc.	TITE
Optimal bucket	0.609	–	0.703	–	0.683	–	0.743	–	0.730	–
Full Thinking	0.559	–	0.611	–	0.569	–	0.635	–	0.540	–
No thinking	0.525	–	0.536	–	0.524	–	0.587	–	0.494	–
Statistical EE	0.560	0.161	0.609	0.060	0.573	0.003	0.635	0.006	0.540	0.045
EntroCut	0.556	0.144	0.595	0.096	0.459	0.000	0.585	0.004	0.563	0.114

Table 5: Average accuracy and TITE on the hallucination benchmarks.

Figure 5: Accuracy versus total generated tokens across five reasoning models. The dashed line indicates full-reasoning accuracy, while the shaded region represents lower-cost operating points with at least the same accuracy. Stars indicate the discrete-budget oracle, and arrows illustrate the remaining oracle gap.

Benchmark	Gemma4 E4B	gpt-oss-20b	Qwen3.5-4B	Qwen3.5-9B	R1-Qwen3-8B
AIME26	9.63	0.83	3.97	3.99	1.40
GPQA-Diamond	19.46	13.33	6.10	5.95	17.34
GSM8K	6.76	1.36	5.66	7.80	1.60
HaluEval	3.13	6.57	6.80	4.97	20.59
LiveCodeBench	8.60	9.66	9.26	8.28	17.90
MATH500	5.22	3.62	2.94	3.74	1.57

Table 6: Continuation reasoning-error rate (CER, %). Lower is better.

Figure 6: Visualization of random exit position and exit traps

Figure 7: (a) Averaged over trajectories. (b) Per trajectory, the observed monotonicity discrepancy

D_{\mathrm{obs}}

exceeds the value expected under a monotone model with decoding noise

D_{\mathrm{null}}

in every setting. The right column gives the excess reversal magnitude

A_{\mathrm{excess}}

Figure 8: Exit traps. (a, b) Share of each method’s exits that fail because of each trap. (c) Each point is one (method, trap) pair: trap prevalence among the method’s exits versus the failure rate on trapped traces. Dashed curves are iso-lines of total failure.

Figure 9: Post-stop reasoning leakage rate by normalized exit position

R_{\mathrm{exit}}

, split into primary leakage and additional soft leakage. Bin width is 0.05.

Model	Method	GPQA-Diamond		LiveCodeBench		MMLU-Pro		TruthfulQA
		$J_{\mathrm{correct}}$	$J_{\mathrm{incorrect}}$	$J_{\mathrm{correct}}$	$J_{\mathrm{incorrect}}$	$J_{\mathrm{correct}}$	$J_{\mathrm{incorrect}}$	$J_{\mathrm{correct}}$	$J_{\mathrm{incorrect}}$
gpt-oss-20b	Full Thinking	85.2	43.6	88.4	77.2	89.4	53.0	88.8	73.4
	EntroCut	-16.3	-20.0	-8.2	-12.1	-12.6	-16.2	-15.3	-23.6
	DTSR-1	-15.0	-20.4	-9.9	-13.4	-11.2	-21.3	-10.0	-18.2
	DEER	-18.9	-19.7	-5.6	-8.8	-10.3	-19.2	-9.8	-26.4
	TERMINATOR	-59.9	-29.2	-8.3	-13.6	-41.4	-38.0	-60.1	-50.7

Table 7: Justification quality by outcome correctness, relative to Full Thinking on the same model and benchmark. Full Thinking rows report the mean judge score (

10

–

100

); other rows report the difference from Full Thinking.

Method	Insuff.	Unsupp.	Circ.	Contra.	Other
Full Thinking	36.1	9.3	2.9	49.0	2.3
DEER	38.8	3.3	1.4	54.4	1.8
EntroCut	44.7	4.1	1.4	48.0	1.3
DTSR-1	52.2	1.5	0.9	44.6	0.7
TERMINATOR	71.7	1.1	0.5	26.1	0.5

Table 8: Failure-mode distribution among low-quality outputs (

J<50

), pooled across all models and benchmarks. Percentages are of

n_{\text{below}}

for that method.

Early Stopping for Language Model Reasoning

See all 26 papers

Early Stopping for Large Reasoning Models via Confidence Dynamics
Apr 6, 2026Parsa Hosseini, Sumit Nawathe, Mahdi Salmani +2Early Stopping for Language Model Reasoning Early Stopping
University of Maryland · University of Southern California
Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models
May 17, 2026Dehai Min, Giovanni Vaccarino, Huiyi Chen +3Large Reasoning Models CoT Reasoning
University of Illinois Chicago · Politecnico di Milano · University of Illinois Urbana-Champaign +1
Thinking Past the Answer: Evaluating Harmful Overthinking in Large Reasoning Models
Jun 1, 2026Simone Caldarella, Davide Talon, Rahaf Aljundi +2LLM Evaluation Overthinking in Language Models
University of Trento · 3Fondazione Bruno Kessler · 2Toyota Motor Europe

Apr 6, 2026J/K move · Enter open · S save

Early Stopping for Large Reasoning Models via Confidence Dynamics

Parsa Hosseini, Sumit Nawathe, Mahdi Salmani, Meisam Razaviyayn+1

University of Maryland · University of Southern California

Early Stopping for Language Model Reasoning Early Stopping Efficient Language Model Reasoning Large Reasoning Models Overthinking in Language Models Confidence Estimation in Language Models

Open paper AskPDF arXiv

Large reasoning models rely on long chain-of-thought generation to solve complex problems, but extended reasoning often incurs substantial computational cost and can even degrade performance due to overthinking. A key challenge is determining when the model should stop reasoning and produce the final answer. In this work, we study the confidence of intermediate answers during reasoning and observe two characteristic behaviors: correct reasoning trajectories often reach high-confidence answers early, while incorrect rollouts tend to produce long, unproductive reasoning traces and exhibit less reliable confidence dynamics. Motivated by these observations, we propose CoDE-Stop (Confidence Dynamics Early Stop), an early stopping method that leverages the dynamics of intermediate answer confidence to decide when to terminate reasoning, requiring no additional training and easily integrating into existing models. We evaluate CoDE-Stop on diverse reasoning and science benchmarks across multiple models. Compared to prior early stopping methods, it achieves a more favorable accuracy-compute tradeoff and reduces total token usage by 25-50% compared to standard full-length reasoning. In addition, we provide analyses of confidence dynamics during reasoning, offering insights into how confidence changes in both correct and incorrect trajectories.

Abstract

Early Stopping for Language Model Reasoning

Explore similar work

Early Stopping for Large Reasoning Models via Confidence Dynamics

Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models

Thinking Past the Answer: Evaluating Harmful Overthinking in Large Reasoning Models