Leaderboard · LiveCodeBench

代码基准榜

LiveCodeBench · 防污染 pass@1 · 共 28 个模型 · 11 月 19 日更新

学界权威的防污染代码基准:题目来自 LeetCode/AtCoder/Codeforces 并带发布日期,只用模型训练截止之后的新题评测,排除背题水分。数据来自 livecodebench.github.io。

01 O4-Mini (High) LiveCodeBench pass@1 · 1055 题 87.3%pass@1
# 名称 变化 pass@1
2 O3 (High) LiveCodeBench pass@1 · 1055 题 – 84.7%
3 O4-Mini (Medium) LiveCodeBench pass@1 · 1055 题 – 84.5%
4 DeepSeek-R1-0528 LiveCodeBench pass@1 · 1055 题 – 84.4%
5 Gemini-2.5-Pro-06-05 LiveCodeBench pass@1 · 1055 题 – 84.3%
6 Gemini-2.5-Pro-05-06 LiveCodeBench pass@1 · 1055 题 – 82.7%
7 OpenReasoning-Nemotron-32B LiveCodeBench pass@1 · 1055 题 – 81%
8 EXAONE-4.0-32B LiveCodeBench pass@1 · 1055 题 – 80.9%
9 Qwen3-235B-A22B LiveCodeBench pass@1 · 1055 题 – 80.4%
10 XBai-o4-medium LiveCodeBench pass@1 · 1055 题 – 80.1%
11 OpenCodeReasoning-Nemotron-1.1-32B LiveCodeBench pass@1 · 1055 题 – 78.5%
12 Grok-3-Mini (High) LiveCodeBench pass@1 · 1055 题 – 78.1%
13 O3-Mini-2025-01-31 (High) LiveCodeBench pass@1 · 1055 题 – 77.7%
14 O4-Mini (Low) LiveCodeBench pass@1 · 1055 题 – 77.4%
15 Gemini-2.5-Flash-05-20 LiveCodeBench pass@1 · 1055 题 – 76.2%
16 O3-Mini-2025-01-31 (Med) LiveCodeBench pass@1 · 1055 题 – 75.4%
17 Gemini-2.5-Flash-04-17 LiveCodeBench pass@1 · 1055 题 – 75.1%
18 O3-Mini-2025-01-31 (Low) LiveCodeBench pass@1 · 1055 题 – 70.6%
19 Claude-Opus-4 (Thinking) LiveCodeBench pass@1 · 1055 题 – 70.4%
20 Claude-Sonnet-4 (Thinking) LiveCodeBench pass@1 · 1055 题 – 68.5%
21 Claude-Opus-4 LiveCodeBench pass@1 · 1055 题 – 62.4%
22 Claude-Sonnet-4 LiveCodeBench pass@1 · 1055 题 – 59.4%
23 DeepSeek-V3 LiveCodeBench pass@1 · 1055 题 – 49.6%
24 Claude-3.5-Sonnet-20241022 LiveCodeBench pass@1 · 1055 题 – 48.7%
25 GPT-4O-2024-08-06 LiveCodeBench pass@1 · 1055 题 – 38.3%
26 GPT-4-Turbo-2024-04-09 LiveCodeBench pass@1 · 1055 题 – 37.3%
27 GPT-4O-mini-2024-07-18 LiveCodeBench pass@1 · 1055 题 – 35.5%
28 Claude-3-Haiku LiveCodeBench pass@1 · 1055 题 – 22.5%