Skip to content

Benchmark

LiveCodeBench Pro

Coding benchmarks compare models on programming tasks. Scores are Elo ratings. Higher ratings are better.

Models measured
38
Scores recorded
42
Sources
3
IQ weight
1.58

Score distribution

1,407
Mean
1,270
Median
787.7
Std dev
402
10th percentile
2,439
90th percentile
0.8
Min
0.82,887
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.58 for chat models, measured over 38 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 3.1 Pro Previewsetting not named by the source· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.2,887
2Gemini 3.1 Pro Preview· LiveCodeBench Pro data
Values from 2 sources
2,887
3Gemini 3 Pro Preview· LiveCodeBench Pro data
Values from 2 sources
2,439
4Gemini 3 Pro PreviewGemini 3 Pro Model Card · results as of November 2025 · Codeforces, ICPC and IOI· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.2,439
5GPT 5.2· LiveCodeBench Pro data2,393
6Gemini 3 Flash· LiveCodeBench Pro data
Values from 2 sources
2,316
7Gemini 3 Flashpublic leaderboard· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.2,316
8GPT 5.1high· LiveCodeBench Pro data2,269
9GPT 5high· LiveCodeBench Pro data2,176
10o4 Mini· LiveCodeBench Pro data2,092
11Doubao Seed 1.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim1,930
12Gemini 2.5 Pro· LiveCodeBench Pro data1,694
13o3 Mini· LiveCodeBench Pro data1,688
14Qwen 3 235B A22B Thinking 2507· LiveCodeBench Pro data1,673
15Qwen3 Next 80B A3Bthinking· LiveCodeBench Pro data1,603
16Claude Sonnet 4.5thinking· LiveCodeBench Pro data1,412
17gemini-2.5-flash-preview-04-17· LiveCodeBench Pro data1,334
18GPT OSS 120B· LiveCodeBench Pro data1,299
19Gemini 2.5 Flash· LiveCodeBench Pro data1,288
20Doubao-Seed 1.6· LiveCodeBench Pro data1,270
21Qwen3 Max· LiveCodeBench Pro data1,226
22Qwen3 30B A3B· LiveCodeBench Pro data1,206
23DeepSeek R1· LiveCodeBench Pro data1,161
24GPT OSS 20b· LiveCodeBench Pro data1,030
25Qwen 3 235B A22B· LiveCodeBench Pro data1,012
26OpenAI o3high· LiveCodeBench Pro data1,010
27QwQ 32B· LiveCodeBench Pro data981
28Grok 4 Fastreasoning· LiveCodeBench Pro data962
29GLM 4.5· LiveCodeBench Pro data941
30GPT-4.1 mini· LiveCodeBench Pro data875
31GPT-4.5 Preview· LiveCodeBench Pro data864
32Grok 4 Fastnon-reasoning· LiveCodeBench Pro data770
33DeepSeek V3· LiveCodeBench Pro data672
34Claude Sonnet 3.7reasoning-disabled· LiveCodeBench Pro data671
35GPT 4.1· LiveCodeBench Pro data606
36Claude 3.5 Sonnet· LiveCodeBench Pro data572
37Llama 4 Maverick· LiveCodeBench Pro data528
38Gemma 3 27B IT· LiveCodeBench Pro data402
39DeepSeek R1 Distill Llama 70b· LiveCodeBench Pro data293
40Qwen Max· LiveCodeBench Pro data274
41GPT 4o· LiveCodeBench Pro data210
42Muse Spark· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim0.8

Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.