Skip to content

Benchmark

LiveCodeBench LeetCode

Coding benchmarks compare models on programming tasks. Scores are percentages.

Models measured
16
Scores recorded
22
Sources
1
IQ weight
2.69

Score distribution

74%
Mean
83.1%
Median
20.3%
Std dev
39%
10th percentile
91.9%
90th percentile
20.7%
Min
20.7%92.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.69 for chat models, measured over 16 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1OpenAI o3platform:leetcode, high· LiveCodeBench Leaderboard data92.8%
2o4 Miniplatform:leetcode, high· LiveCodeBench Leaderboard data92.6%
3o4 Miniplatform:leetcode, medium· LiveCodeBench Leaderboard data91.9%
4Gemini 2.5 Proplatform:leetcode· LiveCodeBench Leaderboard data90.8%
5o3 Miniplatform:leetcode, high· LiveCodeBench Leaderboard data90.8%
6DeepSeek R1platform:leetcode· LiveCodeBench Leaderboard data89.3%
7Qwen 3 235B A22Bplatform:leetcode· LiveCodeBench Leaderboard data87.4%
8o3 Miniplatform:leetcode, Med· LiveCodeBench Leaderboard data86.9%
9Grok 3 Miniplatform:leetcode, high· LiveCodeBench Leaderboard data85.8%
10o4 Miniplatform:leetcode, low· LiveCodeBench Leaderboard data83.3%
11Gemini 2.5 Flashplatform:leetcode· LiveCodeBench Leaderboard data83.1%
12Claude Opus 4platform:leetcode, thinking· LiveCodeBench Leaderboard data80.9%
13o3 Miniplatform:leetcode, low· LiveCodeBench Leaderboard data77.5%
14Claude Sonnet 4platform:leetcode, thinking· LiveCodeBench Leaderboard data77.3%
15Claude Opus 4platform:leetcode· LiveCodeBench Leaderboard data74.5%
16Claude Sonnet 4platform:leetcode· LiveCodeBench Leaderboard data68.5%
17DeepSeek V3platform:leetcode· LiveCodeBench Leaderboard data56.2%
18Claude 3.5 Sonnetplatform:leetcode· LiveCodeBench Leaderboard data55.9%
19GPT 4oplatform:leetcode· LiveCodeBench Leaderboard data45.1%
20GPT-4 Turboplatform:leetcode· LiveCodeBench Leaderboard data39%
21GPT-4o miniplatform:leetcode· LiveCodeBench Leaderboard data35.9%
22Claude 3 Haikuplatform:leetcode· LiveCodeBench Leaderboard data20.7%