REST: Stress Testing Large Reasoning Models by Asking Multiple Problems at Once

TL;DR

REST introduces a stress-testing framework for large reasoning models by evaluating their performance on multiple problems simultaneously, revealing weaknesses not apparent in traditional single-question benchmarks.

Contribution

This work presents REST, a novel multi-problem stress-testing framework that better assesses reasoning models' robustness and capabilities under realistic, multi-context conditions.

Findings

State-of-the-art models degrade significantly under REST stress tests.

REST outperforms existing benchmarks in discriminative power.

Models trained with 'long2short' maintain better performance under stress.

Abstract

Recent Large Reasoning Models (LRMs) have achieved remarkable progress on task-specific benchmarks, yet their evaluation methods remain constrained by isolated problem-solving paradigms. Existing benchmarks predominantly assess single-question reasoning through sequential testing, resulting critical limitations: (1) vulnerability to data contamination and less challenging (e.g., DeepSeek-R1 achieves 97.0% on MATH500), forcing costly creation of new questions with large human efforts, (2) failure to evaluate models under multi-context pressure, a key requirement for real-world deployment. To bridge this gap, we present REST (Reasoning Evaluation through Simultaneous Testing), a stress-testing framework that exposes LRMs to multiple problems simultaneously. Beyond basic reasoning, REST evaluates several under-tested capabilities: contextual priority allocation, cross-problem interference resistance, and dynamic cognitive load management. Our evaluation reveals several striking findings: Even state-of-the-art (SOTA) models like DeepSeek-R1 exhibit substantial performance degradation under stress testing. Crucially, REST demonstrates stronger discriminative power than existing benchmarks, revealing pronounced performance differences among models that exhibit similar, near-ceiling performance under single-question evaluations. Some key insights emerge from our analysis: (1) the "overthinking trap" is a critical factor contributing to the performance degradation; (2) the models trained with "long2short" technique preserve more accuracy of their single-problem performance under REST, outperforming standard-trained counterparts. These results establish REST as a cost-efficient, future-proof evaluation paradigm that better reflects real-world reasoning demands while reducing reliance on continuous human annotation. Code and results are available at https://opendatalab.github.io/REST.