Eval with Claude Computer Use

[Edward-JianQAQ]

Hi,

Thx for the nice work. May I ask for your code to eval the Claude Computer Use setting as I saw they are included in the leaderboard? Thx!

[likaixin2000]

Hi, thanks for your kind words. The Claude results are sourced from UI-TARS, as we don't have Claude API :(

You can reference to this PR which contains Claude support, but we don't have the chance to verify it yet.