TurnBench-MS: A Benchmark for Evaluating Multi-Turn, Multi-Step Reasoning in Large Language Models

TL;DR

TurnBench-MS introduces a comprehensive benchmark for evaluating large language models' ability to perform multi-turn, multi-step reasoning through an interactive code-breaking game, highlighting current models' limitations in complex reasoning tasks.

Contribution

This paper presents TurnBench-MS, a novel benchmark with interactive, multi-turn tasks and annotated reasoning steps, designed to assess and improve multi-step reasoning in large language models.

Findings

State-of-the-art LLMs achieve 84% in Classic mode but only 18% in Nightmare mode.

Humans achieve 100% accuracy in both modes.

TurnBench-MS reveals significant gaps in current models' reasoning capabilities.

Abstract

Despite impressive advances in large language models (LLMs), existing benchmarks often focus on single-turn or single-step tasks, failing to capture the kind of iterative reasoning required in real-world settings. To address this limitation, we introduce TurnBench, a novel benchmark that evaluates multi-turn, multi-step reasoning through an interactive code-breaking task inspired by the "Turing Machine Board Game." In each episode, a model must uncover hidden logical or arithmetic rules by making sequential guesses, receiving structured feedback, and integrating clues across multiple rounds. This dynamic setup requires models to reason over time, adapt based on past information, and maintain consistency across steps-capabilities underexplored in current benchmarks. TurnBench includes two modes: Classic, which tests standard reasoning, and Nightmare, which introduces increased complexity and requires robust inferential chains. To support fine-grained analysis, we provide ground-truth annotations for intermediate reasoning steps. Our evaluation of state-of-the-art LLMs reveals significant gaps: the best model achieves 84% accuracy in Classic mode, but performance drops to 18% in Nightmare mode. In contrast, human participants achieve 100% in both, underscoring the challenge TurnBench poses to current models. By incorporating feedback loops and hiding task rules, TurnBench reduces contamination risks and provides a rigorous testbed for diagnosing and advancing multi-step, multi-turn reasoning in LLMs.