上下文窗口1.1M
最大输出128K
知识截止2026-02-16
发布日期2026-07-09
系列gpt-luna
最后更新2026-07-09
输入模态文本图像PDF
输出模态文本
能力
推理工具调用结构化输出文件附件
基准跑分
SWE-Bench Pro 62.7 (resolve rate) · Terminal-Bench 84.7 (success rate) · DeepSWE 67.2 (resolve rate) · GPQA Diamond 92.3 (accuracy) · FrontierMath 78.6 (accuracy) · BrowseComp 83.3 (accuracy) · OSWorld 45.6 (success rate) · MMMU Pro 78.4 (accuracy) · Agents' Last Exam 50.3 · Toolathlon 53.4 (success rate) · Artificial Analysis Intelligence Index 51.2 (index score) · Artificial Analysis Coding Agent Index 74.6 (index score)