LiveCodeBench

text
+
+
+
+
About

LiveCodeBench is a holistic and contamination-free code evaluation benchmark that continuously collects new programming problems from competitive coding platforms. This dynamic benchmark tests AI models' programming capabilities through fresh problems that evolve over time, preventing memorization and ensuring genuine coding skills assessment across algorithm implementation, problem-solving, and code generation tasks.

+
+
+
+
Evaluation Stats
Total Models50
Organizations10
Verified Results0
Self-Reported50
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers

Score Distribution

50 models
Top Score
80.4%
Average Score
47.7%
High Performers (80%+)
2

Top Organizations

#1xAI
5 models
79.6%
#2Zhipu AI
2 models
71.8%
#3NVIDIA
2 models
68.7%
#4Moonshot AI
1 model
53.7%
#5Microsoft
2 models
53.4%
+
+
+
+
Leaderboard
50 models ranked by performance on LiveCodeBench
LicenseLinks
Feb 17, 2025
Proprietary
80.4%
Aug 28, 2025
Proprietary
80.0%
Jul 9, 2025
Proprietary
79.4%
Feb 17, 2025
Proprietary
79.4%
Jul 9, 2025
Proprietary
79.0%
Sep 29, 2025
MIT
74.1%
May 28, 2025
MIT
73.3%
Jul 28, 2025
MIT
72.9%
Aug 18, 2025
NVIDIA Open Model License Agreement
71.1%
Jul 28, 2025
MIT
70.7%
Showing 1 to 10 of 50 models
...
+
+
+
+
Resources