LiveCodeBench-CPP is a C++ code generation benchmark dataset released by NVIDIA in 2025, designed to provide a benchmark for evaluating the code generation and debugging capabilities of large language models based on the latest programming problems.

The dataset contains 454 programming problems, primarily sourced from AtCoder (287 problems) and LeetCode (167 problems), covering contests and problems from October 2024 to May 2025. The data includes problem descriptions, platform sources, difficulty levels, test cases, and related metadata, primarily used to evaluate models' code generation and problem-solving abilities in real-world programming tasks.

Dataset Composition

• question_title: The title of the programming problem.
• question_content: The problem description.
• platform: The source platform of the problem, including AtCoder and LeetCode.
• question_id: The unique identifier of the problem.
• contest_id: The unique identifier of the contest.
• contest_date: The date of the contest.
• starter_code: The code template or function signature.
• difficulty: The difficulty level of the problem.
• public_test_cases: Public test cases.
• private_test_cases: Private test cases.
• metadata: Metadata related to the problem.