# How Clued up are LLMs? Evaluating Multi-Step Deductive Reasoning in a Text-Based Game Environment

arXiv: https://arxiv.org/abs/2603.17169 · HTML: https://mlsanity.com/paper/2603.17169/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 20 of its 26 references that are in the map.
- First posted on arXiv 17 Mar 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 20 of 20)

- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [ReClor: A Reading Comprehension Dataset Requiring Logical Reasoning](https://mlsanity.com/paper/2002.04326/) (2020)
- [Navigate through Enigmatic Labyrinth A Survey of Chain of Thought Reasoning: Advances, Frontiers and Future](https://mlsanity.com/paper/2309.15402/) (2023)
- [Language Agents with Reinforcement Learning for Strategic Play in the Werewolf Game](https://mlsanity.com/paper/2310.18940/) (2023)
- [Evaluating Mathematical Reasoning Beyond Accuracy](https://mlsanity.com/paper/2404.05692/) (2024)
- [MR-GSM8K: A Meta-Reasoning Benchmark for Large Language Model Evaluation](https://mlsanity.com/paper/2312.17080/) (2023)
- [Multi-LogiEval: Towards Evaluating Multi-Step Logical Reasoning Ability of Large Language Models](https://mlsanity.com/paper/2406.17169/) (2024)
- [On the Impact of Fine-Tuning on Chain-of-Thought Reasoning](https://mlsanity.com/paper/2411.15382/) (2024)
- [The Curse of CoT: On the Limitations of Chain-of-Thought in In-Context Learning](https://mlsanity.com/paper/2504.05081/) (2025)
- [AMONGAGENTS: Evaluating Large Language Models in the Interactive Text-Based Social Deduction Game](https://mlsanity.com/paper/2407.16521/) (2024)
- [Probing the Multi-turn Planning Capabilities of LLMs via 20 Question Games](https://mlsanity.com/paper/2310.01468/) (2023)
- [DyFlow: Dynamic Workflow Framework for Agentic Reasoning](https://mlsanity.com/paper/2509.26062/) (2025)
- [PokerBench: Training Large Language Models to become Professional Poker Players](https://mlsanity.com/paper/2501.08328/) (2025)
- [GAMEBoT: Transparent Assessment of LLM Reasoning in Games](https://mlsanity.com/paper/2412.13602/) (2024)
- [Exploring the Necessity of Reasoning in LLM-based Agent Scenarios](https://mlsanity.com/paper/2503.11074/) (2025)
- [Practical Considerations for Agentic LLM Systems](https://mlsanity.com/paper/2412.04093/) (2024)
- [Beyond Correctness: Rewarding Faithful Reasoning in Retrieval-Augmented Generation](https://mlsanity.com/paper/2510.13272/) (2025)

## Built on by (most cited first, 1 of 1)

- [Clueing up LLMs with Tool-Augmented Deductive Reasoning](https://mlsanity.com/paper/2609.18736/) (2026)

## Abstract (arXiv)

Deducing whodunit proves challenging for LLM agents. In this paper, we implement a text-based multi-agent version of the classic board game Clue as a rule-based testbed for evaluating multi-step deductive reasoning, with six agents drawn from GPT-4o-mini and Gemini-2.5-Flash. We further investigate whether fine-tuning on structured logic puzzles transfers to improved in-game reasoning and gameplay. Across 18 simulated games, agents achieve only four correct wins, indicating difficulty in maintaining consistent deductive reasoning over the course of a full game. Additionally, we find that fine-tuning does not reliably improve performance and, in some cases, appears to increase reasoning volume without improving reasoning precision.