Multi-LCB: Extending LiveCodeBench to Multiple Programming Languages
Summary
Multi-LCB extends the LiveCodeBench benchmark to evaluate LLMs across twelve programming languages while preserving contamination controls, revealing Python overfitting and language-specific contamination issues.
View Cached Full Text
Cached at: 06/20/26, 02:27 PM
Paper page - Multi-LCB: Extending LiveCodeBench to Multiple Programming Languages
Source: https://huggingface.co/papers/2606.20517
Abstract
Multi-LCB addresses the limitation of LiveCodeBench by providing a multi-language benchmark for evaluating LLMs across twelve programming languages while maintaining contamination controls and evaluation protocols.
LiveCodeBench (LCB) has recently become a widely adopted benchmark for evaluatinglarge language models(LLMs) oncode-generation tasks. By curatingcompetitive programming problems, constantly adding fresh problems to the set, and filtering them by release dates, LCB providescontamination-aware evaluationand offers a holistic view of coding capability. However, LCB remains restricted to Python, leaving open the question of whether LLMs can generalize across the diverse programming languages required in real-world software engineering. We introduce Multi-LCB, a benchmark for evaluating LLMs across twelve programming languages, including Python. Multi-LCB transforms Python tasks from the LCB dataset into equivalent tasks in other languages while preserving LCB’s contamination controls and evaluation protocol. Because it is fully compatible with the original LCB format, Multi-LCB will automatically track future LCB updates, enabling systematic assessment ofcross-language code generationcompetence and requiring models to sustain performance well beyond Python. We evaluated 24 LLMs for instruction and reasoning on Multi-LCB, uncovering evidence ofPython overfitting,language-specific contamination, and substantial disparities inmultilingual performance. Our results establish Multi-LCB as a rigorous new benchmark for multi-programming-language code evaluation, directly addressing LCB’s primary limitation and exposing critical gaps in current LLM capabilities.
View arXiv pageView PDFProject pageGitHub22Add to collection
Get this paper in your agent:
hf papers read 2606\.20517
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.20517 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.20517 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.20517 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
PolyWorkBench: Benchmarking Multilingual Long-Horizon LLM Agents
Introduces PolyWorkBench, a benchmark for evaluating LLM agents on multilingual long-horizon workplace workflows across five domains, demonstrating significant performance degradation compared to monolingual settings.
Terminal-Bench-LILT: Multilingual Agentic Coding Benchmark Grounded in Language, Region, and Culture
Terminal-Bench-LILT presents a multilingual coding benchmark with 300 authentic tasks in 10 languages, exposing gaps in AI models' handling of non-English issues like internationalization and cultural conventions.
Benchmarking LLMs
A study or report on benchmarking large language models, likely comparing performance across various tasks.
ParBench: A Benchmark for Reliable Evaluation of LLM Parallel Code Translation
ParBench is a benchmark framework for evaluating LLM-based parallel API translation across CUDA, OpenMP, OpenCL, and OpenMP target offload, focusing on executable, reproducible conditions and robustness testing.
Open-source LLM benchmark runs 147 coding tasks every 4 hours, 5-trial median with 95% CI, and uses CUSUM for change-point detection. Curious what people think of the methodology
An open-source LLM benchmark with 147 coding tasks runs every 4 hours, using 5-trial median with 95% confidence intervals and CUSUM for change-point detection, sparking discussion on its methodology.