EvoEval Combine — leaderboard

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

Top models

#ModelScore
1GPT-453
2GPT-4 Turbo45
3Claude 3 Haiku25
4Gemini 1.0 Pro23
5Claude 219
6deepseek-coder-6.7B-instruct18
7Phi-214
8Mixtral 8x7B Instruct9
9deepseek-coder-1.3B-instruct9
10starcoder2-15B5
11Mistral 7B Instruct3
12Mistral 7B3
13gemma-7B2
14gemma-2B2
15starcoder2

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