Skip to content

Benchmark

HumanEval

Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.

Models and systems measured
98
Scores recorded
99
Sources
3
IQ weight
1.81

Score distribution

71.5%
Mean
78.7%
Median
21.1%
Std dev
37.8%
10th percentile
92%
90th percentile
11.6%
Min
11.6%96.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.81 for chat models, measured over 96 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1O1 Mini· EvalPlus Leaderboard data
Values from 2 sources
  • EvalPlus Leaderboard data · 96.3% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
  • LLM Stats data · 92.4% · Data by LLM Stats
96.3%
2O1 Preview· EvalPlus Leaderboard data96.3%
3Kimi K2 0905· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim94.5%
4GPT 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim93.4%
5Kimi K2 Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim93.3%
6GPT 4o· EvalPlus Leaderboard data
Values from 2 sources
  • EvalPlus Leaderboard data · 92.7% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
  • LLM Stats data · 90.2% · Data by LLM Stats
92.7%
7Qwen 2.5 Coder 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
  • LLM Stats data · 92.7% · Data by LLM Stats
  • EvalPlus Leaderboard data · 92.1% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
92.7%
8Sarvam-30B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim92.1%
9Claude 3.5 Sonnet· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim92%
10Mistral Large 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim92%
11DeepSeek V3· EvalPlus Leaderboard data91.5%
12DeepSeek V2.5· EvalPlus Leaderboard data
Values from 2 sources
  • EvalPlus Leaderboard data · 90.2% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
  • LLM Stats data · 89% · Data by LLM Stats
90.2%
13gemini-1.5-pro-002· EvalPlus Leaderboard data89%
14Llama 3.1 405B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim89%
15Nova Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim89%
16LongCat Flash Chat· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
17Mistral Small 3.1 24b Instruct 2503· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
18GPT 4· EvalPlus Leaderboard data
Values from 2 sources
  • EvalPlus Leaderboard data · 88.4% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
  • LLM Stats data · 67% · Data by LLM Stats
88.4%
19Grok 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
20Grok Beta· EvalPlus Leaderboard data88.4%
21Llama 3.3 70B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
22Qwen2.5 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
23Qwen2.5 Coder 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.4%
24Claude Haiku 3.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.1%
25o1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88.1%
26GPT-4.5 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim88%
27Gemma 3 27B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim87.8%
28GPT-4o mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim87.2%
29GPT-4 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
  • LLM Stats data · 87.1% · Data by LLM Stats
  • EvalPlus Leaderboard data · 85.4% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
87.1%
30Qwen2.5 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim86.6%
31Qwen2 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim86%
32Grok 2 mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim85.7%
33DeepSeek Coder V2 Instruct· EvalPlus Leaderboard data85.4%
34Gemma 3 12B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim85.4%
35Nova Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim85.4%
36Claude 3 Opus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
  • LLM Stats data · 84.9% · Data by LLM Stats
  • EvalPlus Leaderboard data · 82.9% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
84.9%
37Mistral Small 24B Instruct2501· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim84.8%
38Qwen2.5 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim84.8%
39Gemini 1.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim84.1%
40Qwen 2.5 14B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.5%
41Phi 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim82.6%
42gemini-1.5-flash-002· EvalPlus Leaderboard data82.3%
43DeepSeek Coder 33B Instruct· EvalPlus Leaderboard data81.1%
44Nova Micro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim81.1%
45Llama 3.1 70B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim80.5%
46Llama 3.1 70B Instructofficial instruction-tuned table· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.5%
47Qwen2 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim79.9%
48Qwen2.5-Omni 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim78.7%
49Llama 3 70B Instruct· EvalPlus Leaderboard data77.4%
50Claude 3 Haiku· EvalPlus Leaderboard data
Values from 2 sources
  • EvalPlus Leaderboard data · 76.8% · EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages
  • LLM Stats data · 75.9% · Data by LLM Stats
76.8%

Showing 1–50 of 98 results.

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
DeepSeek V4.1 Flash Basepass@1 · 0-shot· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.79.4%

Data credits: EvalPlus Leaderboard and EvalPerf (evalplus.github.io), Apache-2.0; EvalPerf win-rate fractions converted to percentages (licence) · 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.