CresOWLve: Benchmarking Creative Problem-Solving Over Real-World Knowledge

TL;DR

CresOWLve is a new benchmark designed to evaluate large language models' ability to solve real-world, creative problems that require integrating diverse knowledge and thinking strategies.

Contribution

It introduces a realistic, knowledge-grounded benchmark for assessing creative problem-solving in LLMs, addressing limitations of previous artificial benchmarks.

Findings

Models perform better on factual questions than creative ones, with up to 17% performance drop.

Models can retrieve relevant knowledge but struggle with forming creative connections.

CresOWLve remains highly challenging for current LLMs.

Abstract

Creative problem-solving requires combining multiple cognitive abilities, including logical reasoning, lateral thinking, analogy-making, and commonsense knowledge, to discover insights that connect seemingly unrelated pieces of information. However, most existing benchmarks for large language models (LLMs) evaluate only specific components of this process. Moreover, many creativity-oriented benchmarks rely on artificially constructed brainteasers or contrived scenarios that do not reflect how creative problem-solving occurs in real-world settings. To address this gap, we introduce CresOWLve, a benchmark for evaluating creative problem-solving using puzzles grounded in real-world knowledge. Problems in CresOWLve require employing multiple creative thinking strategies, retrieving facts from diverse domains, and creatively combining them to arrive at a solution. Evaluating several frontier non-thinking and thinking LLMs, we show that CresOWLve remains highly challenging. Our analysis reveals a consistent performance gap: models perform substantially better on factual questions than on creative ones (up to a -17% drop). While models can often retrieve the relevant knowledge, they struggle to form the non-obvious creative connections required to integrate this information and arrive at the correct answer.