EvoEval Verbose — leaderboard

Metric: Pass@1 (%). Source: evo-eval.github.io. 17 models tracked.

Top models

#ModelScore
1GPT-4 Turbo79.27
2deepseek-coder-6.7B-instruct72.56
3Claude 3 Haiku71.34
4GPT-470.12
5Claude 266.46
6Gemini 1.0 Pro58.54
7deepseek-coder-1.3B-instruct57.32
8Phi-251.22
9starcoder2-15B46.34
10Mixtral 8x7B Instruct40.85
11starcoder2-7B37.8
12starcoder34.76
13gemma-7B33.54
14starcoder2-3B32.93
15Mistral 7B31.1

Interactive version: aibenchmarks.dev/benchmark?slug=evoeval-verbose · How the rankings work · Data refreshed daily, snapshot 2026-07-08.