EfficientXLang: Towards Improving Token Efficiency Through Cross-Lingual Reasoning
On the map: LLM reasoning → Reasoning benchmarks and evaluation
Abstract (arXiv)
Despite recent advances in Language Reasoning Models (LRMs), most research focuses solely on English, even though many models are pretrained on multilingual data. In this work, we investigate: Is English the most token-efficient language for reasoning? We evaluate three open-source RLMs: DeepSeek R1, Qwen 2.5 and Qwen 3, across four math datasets and seven typologically diverse languages. We find that reasoning in non-English languages not only reduces token usage, but also preserves accuracy. These gains persist even after translating the reasoning traces into English, suggesting genuine shifts in reasoning behavior rather than surface-level linguistic effects. The extent of improvement, however, depends on the models multilingual strength. Our findings motivate a broader view of reasoning in language models, highlighting the potential of multilingual reasoning and the importance of strong multilingual foundations. The code for our work can be found: https://github.com/microsoft/EfficientXLang.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 18 of its 31 references that are in the map. First posted on arXiv 30 Jun 2025.
Builds on
18 of 31 references.
Closest ancestors · 4
Same direction, newest first
- 2025Multilingual Test-Time Scaling via Initial Thought Transfer21 May
- 2025Could Thinking Multilingually Empower LLM Reasoning?16 Apr
- 2025DeepSeek-R1 Thoughtology: Let's think about LLM Reasoning2 Apr
- 2025DNR Bench: Benchmarking Over-Reasoning in Reasoning LLMs20 Mar
Other references · 14
- 2021Evaluating Large Language Models Trained on Codecited by 8,320
- 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
- 2024Qwen2.5 Technical Reportcited by 4,942
- 2024GPT-4o System Cardcited by 4,588
- 2023Let's Verify Step by Stepcited by 3,789
- 2024OpenAI o1 System Cardcited by 1,244
- 2016Bag of Tricks for Efficient Text Classificationcited by 876
- 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMscited by 570
- 2022Language Models are Multilingual Chain-of-Thought Reasonerscited by 530
- 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Modelscited by 464
- 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learningcited by 384
- 2025Reasoning Models Can Be Effective Without Thinkingcited by 187
Built on by
Citing papers per year: 2026: 3.
3 papers.
- 2026Language of Thought Shapes Output Diversity in Large Language Modelscited by 4
- 2026Less Languages, Less Tokens: An Efficient Unified Logic Cross-lingual Chain-of-Thought Reasoning Frameworkcited by 1
- 2026Cross-Lingual Token Arbitrage: Optimizing Code Agent Context Windows via Local LLM Preprocessingcited by 1
- 2026Cross-Lingual Token Arbitrage: Optimizing Code Agent Context Windows via Local LLM Preprocessing2 Jun 2026
- 2026Less Languages, Less Tokens: An Efficient Unified Logic Cross-lingual Chain-of-Thought Reasoning Framework22 Apr 2026
- 2026Language of Thought Shapes Output Diversity in Large Language Models16 Jan 2026
Similar, not cited
Close in wording, no citation either way
- 2025MMATH: A Multilingual Benchmark for Mathematical Reasoning
- 2025Crosslingual Reasoning through Test-Time Scaling
- 2025MathMist: A Parallel Multilingual Benchmark Dataset for Mathematical Problem Solving and Reasoning
- 2026Beyond Input Understanding: Diagnosing Multilingual Mathematical Reasoning with Directed Acyclic Trace Graphs
- 2026Rethinking the Multilingual Reasoning Gap with Layer Swap
- 2026What Makes Good Multilingual Reasoning? Disentangling Traces with Measurable Features
- 2025PolyMath: Evaluating Mathematical Reasoning in Multilingual Contexts
- 2025Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models