跳到正文

GPT Image 1 Mini (medium)

OpenAI · 生图与视频模型资料

采用第三方公开原榜的模型标识及原始成绩。发布日期为来源登记日期,国别、官方渠道和商用许可按已有证据记录,未知项保留待核验。

文生图名次

70

原榜 Elo

935.8

比较次数 · 非人数

2,333

来源登记发布日期

2025-10-06

各任务原榜表现

文生图 →

AA-Image-T2I v2.1 Overall

935.8 Elo · 原榜第70名

95% Elo 区间
918.82–952.82
原榜名次区间
待核验
比较次数 · 非人数
2,333
来源参考价格
$11 /千张
资料快照
2026年10月22日 04:02
原始评测截止日
来源未提供
评测人群、价格条件与来源说明

AA-Image-T2I v2.1 Overall;text-to-image;not-applicable;Artificial Analysis原榜偏好评测,Elo仅在同任务/协议/版本比较。比较次数非人数;原评测截止日及原分数更新时间未提供。价格是第三方参考,不是独立核实的官方报价。开放权重不证明开源/商业许可。来源价格条件原文:API Pricing reflects the cost to generate 1,000 images on the model creator's API at 1024x1024 with the model's default settings。来源评审说明原文(What are the top models on AA-Image-T2I v2.1?):The top models on AA-Image-T2I v2.1 by Elo rating are: 1. GPT Image 2.5 Sunburst (max) (Elo 1222), 2. GPT Image 2.5 Flare (max) (Elo 1207), 3. GPT Image 2 (high) (Elo 1186), 4. Nano Banana 2.1 (Elo 1176), and 5. MAI-Image-2.6 (Elo 1166). Rankings are based on blind votes from our validated private evaluators. image/95a40d42-7e24-42ac-8e86-ded5ba37c6ea: SSR Range 68-75 conflicts with JSON 67–75; rank interval is unverified. 来源storedRank=69不作当前名次;当前SSR、formatted.rank与raw零起下标一致为70。 来源说明:Generated using the OpenAI Images API with `model_id: "gpt-image-1-mini"` and `quality: "medium"`. Prompts were pre-pended with "Create the following as a square image:" for Text to Image, and "Edit the following to create a square image:" for Image Editing. We found without pre-pending these instructions, images would not be produced natively as squares.

各任务和音频模式独立比较;Elo 不转成 FLABS 任务试评分,也不跨榜平均。偏好成绩与价格供选型参考,实际生成效果、输入素材与商用许可仍需结合工作任务核对。