EvoEval Tool Use — leaderboard

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

Top models

#ModelScore
1GPT-4 Turbo69
2GPT-468
3Claude 3 Haiku61
4Gemini 1.0 Pro57
5Claude 257
6deepseek-coder-6.7B-instruct51
7starcoder2-15B48
8starcoder2-7B46
9Mixtral 8x7B Instruct45
10starcoder44
11gemma-7B39
12deepseek-coder-1.3B-instruct39
13Mistral 7B38
14Phi-237
15starcoder2-3B35

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