Artificial Analysis：Nano Banana 2.1 在全部 7 项图像编辑动作上成为 Google 最佳模型
In Image Editing, Nano Banana 2.1 is now Google's best model on all 7 editing actions, with big gains in Scene & Style and Identity Preserving edits
Artificial Analysis 的 Image Arena 评测显示，Nano Banana 2.1 在全部 7 项图像编辑动作上成为 Google 最佳模型，其中 Scene & Style 编辑从第 24 名升至紧随领先模型，Identity Preserving 编辑从第 12 名升至第 3 名。在 Reasoning-Based Edit 和 Identity Preserving Edit 两项上，只有两个 GPT Image 2.5 模型排在它前面。Enhancement & Restoration 是短板，排在第 18 名，与 Nano Banana 2 持平。
推荐理由
精选AI 评分 67
Artificial Analysis 的图像编辑榜单显示 Nano Banana 2.1 在 7 项编辑动作上全面领先，可据此判断 Google 图像编辑能力的当前位次。
主题
正文
当前显示原文，中文全文译文尚未就绪。
In Image Editing, Nano Banana 2.1 is now Google's best model on all 7 editing actions, with big gains in Scene & Style and Identity Preserving edits
Our Image Arena taxonomy measures 7 editing actions and 10 use cases for Image Editing. Editing actions measure what an edit asks the model to do.
Only the two GPT Image 2.5 models rank above Nano Banana 2.1 on Reasoning-Based Edit and Identity Preserving Edit. Scene & Style Edit jumps from #24 to close behind the leading model. Identity Preserving Edit rises from #12 to #3, so it keeps people and characters consistent through an edit better than Nano Banana 2. Enhancement & Restoration is the weak spot: #18, and no better than Nano Banana 2.
➤ Reasoning-Based Edit covers edits that need inference about the scene rather than a literal instruction.
➤ Scene & Style Edit covers relighting, restyling, and changes to the background, setting or overall treatment.
➤ Identity Preserving Edit covers changes to people, faces and characters that must keep identity intact.