测试集:SWE-Bench Verified · 条件:Official announcement; harness, agent scaffold, sampling and evaluation date not separately disclosed. · 评测方:Mistral AI · 日期:未公开
测试集:tau^3-Telecom · 条件:Official announcement; task protocol, tool configuration, sampling and evaluation date not separately disclosed. · 评测方:Mistral AI · 日期:未公开