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-sol
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 64.6 (resolve rate) · Terminal-Bench 88.8 (success rate) · DeepSWE 72.7 (resolve rate) · GPQA Diamond 94.6 (accuracy) · FrontierMath 89 (accuracy) · BrowseComp 90.4 (accuracy) · OSWorld 62.6 (success rate) · MMMU Pro 83 (accuracy) · Agents' Last Exam 52.7 · Toolathlon 58 (success rate) · Artificial Analysis Intelligence Index 58.9 (index score) · Artificial Analysis Coding Agent Index 80 (index score)