Model Releases
More Capable, Less Faithful: A Multilingual Analysis of Mathematical (Un)Solvability Detection in LLMs
arXiv:2608.30463v1 Announce Type: new Abstract: Solvability detection is one of the most challenging aspects of mathematical reasoning for Large Language Models (LLMs). While prior work has studied th
arXiv:2608.30463v1 Announce Type: new Abstract: Solvability detection is one of the most challenging aspects of mathematical reasoning for Large Language Models (LLMs). While prior work has studied this capability extensively, these analyses have been limited to English. Consequently, it remains unclear whether multilingual failures arise from differences in internal Solvability Belief or from language-dependent failures to express it. To address this gap, we introduce the first multilingual benchmark of paired solvable and unsolvable mathematical problems, extending ReliableMath to French and Greek. Using this, we train multilingual probes predicting Solvability Belief and analyze the solvability detection capabilities of state-of-the-art LLMs behaviorally, representationally, and in terms of faithfulness. We find that Solvability Belief is encoded as a largely universal, language-agnostic feature, and that higher-resource languages such as English, despite achieving stronger mathematical reasoning performance, exhibit lower solvability-detection faithfulness.
Related
- Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models
- Not All Proofs Are Equal: Evaluating LLM Proof Quality Beyond Correctness
- SMRC: Aligning Large Language Models with Student Reasoning for Mathematical Error Correction
- AgentCoMa: A Compositional Benchmark Mixing Commonsense and Mathematical Reasoning in Real-World Scenarios
Source: arXiv cs.CL | 2026-09-01