live
weekly refresh
basedagi.org
benchmark evidence

LiveCodeBench

LiveCodeBench coding benchmark scores from LLM-Stats.

winner on LiveCodeBench
direct benchmark result, not a broad vertical composite | source row dated 2000-01-01
scored on 2025-02-01 · stale source data (483d)
latest mapped results | top 20
#ModelScoreEvidenceTested
1DeepSeek: DeepSeek V4 Pro
Deepseek
93.5
model-only
independent_benchmark
2000-01-01
2DeepSeek: DeepSeek V4 Flash
Deepseek
91.6
model-only
independent_benchmark
2000-01-01
3OpenAI: o4 Mini
Openai
74.5
model-only
independent_benchmark
2025-02-01
4OpenAI: o1
Openai
71.0
model-only
independent_benchmark
2025-02-01
5Qwen: Qwen3 235B A22B
Qwen
70.7
model-only
independent_benchmark
2000-01-01
6OpenAI: o3 Mini
Openai
69.5
model-only
independent_benchmark
2025-02-01
7Google: Gemini 2.5 Pro
Google
69.2
model-only
independent_benchmark
2025-02-01
8Qwen: Qwen3 32B
Qwen
65.7
model-only
independent_benchmark
2000-01-01
9DeepSeek: R1
Deepseek
62.8
model-only
independent_benchmark
2025-02-01
10Qwen: Qwen3 30B A3B
Qwen
62.6
model-only
independent_benchmark
2000-01-01
11MoonshotAI: Kimi K2 Thinking
Moonshotai
53.7
model-only
independent_benchmark
2000-01-01
12Meta: Llama 4 Maverick
Meta Llama
43.4
model-only
independent_benchmark
2000-01-01
13DeepSeek: DeepSeek V3.2
Deepseek
34.6
model-only
independent_benchmark
2025-02-01
14OpenAI: GPT-5
Openai
32.9
model-only
independent_benchmark
2025-02-01
15Google: Gemini 2.0 Flash
Google
31.8
model-only
independent_benchmark
2025-02-01
16Mistral Large
Mistralai
27.1
model-only
independent_benchmark
2025-02-01
what this result means

LiveCodeBench coding benchmark scores from LLM-Stats.

This benchmark contributes direct public evidence. Read its scope before generalizing the result.

A win here is a win on LiveCodeBench. Broad task pages require independent corroboration before naming a general winner.

source record
category: coding
metric: pass@1
matched models: 16
latest source date: 2025-02-01
direction: higher is better
inspect upstream source ->