x1: Learning to Think Adaptively Across Languages and Cultures

2026-04-18

AI Summary

Current large language models are typically constrained to reasoning in a single dominant language, overlooking the unique cultural priors and abstract reasoning capacities embedded in different languages. This work proposes the x1 series of models, which— for the first time—formalizes language selection as a functional component within the reasoning process. By employing a multilingual contrastive learning strategy, the model adaptively selects the optimal reasoning language for each input without increasing its knowledge capacity. The approach yields significant performance gains on multilingual mathematical reasoning and culturally grounded tasks. Moreover, the study reveals that while model scaling can mitigate cross-lingual disparities in procedural tasks, specific languages retain irreplaceable advantages in the efficiency and accuracy of knowledge retrieval for culturally sensitive content.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Knowledge Representation Languages

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Languages encode distinct abstractions and inductive priors, yet most large language models (LLMs) overlook this diversity by reasoning in a single dominant language. In this work, we introduce x1, a family of reasoning models that can adaptively reason in an advantageous language on a per-instance basis. To isolate the effect of reasoning-language choice, x1 is constructed without expanding the model's knowledge boundaries and is trained by contrasting linguistically distinct reasoning trajectories for the same input. Our extensive experiments demonstrate the benefits of adaptive multilingual reasoning across multilingual mathematical reasoning and culturally grounded tasks. Moreover, our results challenge a simplistic view of scaling laws: while scaling reduces cross-lingual disparities in procedural domains such as math reasoning, it does not eliminate the advantages of culture-associated languages in culturally grounded tasks, as we empirically show that such reasoning enables more efficient and accurate cultural knowledge recall. Overall, our findings establish language choice as a functional component of reasoning, with implications for building more generalist and globally competent reasoning models.

Problem

• multilingual reasoning
• language choice
• cultural grounding
• inductive priors
• cross-lingual disparities

Innovation

• adaptive multilingual reasoning
• language choice
• culturally grounded tasks
• cross-lingual disparities
• inductive priors