GPT Image 2
预览OpenAI全部 OpenAI 模型
$5.00
$30.00
2026-04-21
定位:旗舰。
- 定位
- 旗舰,依原厂说法。 原厂已推出新一代。来源
- 价格
- 不标:这款没有可比的1K 每张价格。
- 原厂速度
- 不标:原厂没有比较这款的速度。
三项各自独立,不是质量评分。
gpt-image-2 是 OpenAI 的 GPT Image 2 图像生成与编辑模型,2026 年 4 月 21 日开放 API 调用。OpenAI 的提示指南推荐用于文字密集图像、拟真摄影与合成,并支持最高 3840×2160(4K)的自定义尺寸。在 ATP 为文生图 preview 上架,按 token 计费。
模态
图像
文本入
$5.00
USD / 百万 Token
图片出
$30.00
USD / 百万 Token
发布
2026 年 4 月 21 日
价格
用多少付多少:请求成功才扣点,失败不收费。
文本入
$5.00
图片出
$30.00
USD / 百万 Token
文本入
$5.00
500 点数 · USD / 百万 Token
图片出
$30.00
3,000 点数 · USD / 百万 Token
文生图,按 token 计费(文本输入 + 图片输出)。比照原厂牌价直通——无 ATP markup。正式开放前数字可能调整;图片输入与缓存读维度尚未呈现。
示范
全部是模型直接生成的原始结果,没有后期处理。展开任意一条就能看到完整提示词。

提示词
A single unbranded white leather low-top sneaker, laces neatly tied, floating slightly above a seamless pure-white surface with a soft contact shadow beneath it. Studio product shot, three-quarter front angle, 85mm lens, deep focus so the stitching and eyelets stay sharp. Large softbox from upper left with a white bounce card filling the shadow side, clean high-key commercial grade, no color cast, extreme detail, photorealistic.

提示词
An opened plain kraft shipping box on a walnut desk, tissue paper folded back to reveal an unbranded matte-black wireless earbud case, with a small blank thank-you card, a coil of USB-C cable, and a pair of scissors arranged around it. Overhead flat lay, perfectly top-down, 35mm lens, deep focus edge to edge. Diffused overcast daylight from a large window, warm neutral grade, natural wood tones, extreme detail, photorealistic.

提示词
A clear glass latte with distinct milk layering sits on a dark walnut counter, thin steam curling from the surface, a linen napkin and two coffee beans beside it. Background of a blurred cafe interior with warm pendant lights. Medium close-up, slightly below eye level to emphasise the layers, 50mm lens, shallow depth of field. Backlit from a window behind so the steam catches the light, warm tungsten accents against cool ambient shade, rich contrast commercial grade, extreme detail, photorealistic.

提示词
A Taiwanese man in his early 40s, short neat black hair greying at the temples, clean-shaven, calm confident expression, medium build, wearing a charcoal suit without a tie, stands beside a large wall display presenting to four seated colleagues around a light oak table. The display shows only a plain neutral gradient, no text or charts. Glass-walled meeting room with a city skyline blurred beyond. Wide shot, eye level, 35mm lens, deep focus. Soft daylight from the glass wall with even ceiling fill, cool neutral corporate grade, extreme detail, photorealistic. No readable text anywhere in frame.

提示词
Five colleagues of mixed ages and ethnicities stand in a loose semicircle before a large whiteboard covered in blank sticky notes and plain arrows, one woman mid-gesture explaining, the others listening with engaged expressions. Open-plan office with plants and warm wood accents softly out of focus behind. Wide shot, eye level, 35mm lens, moderate depth of field keeping all five readable. Diffused daylight from tall windows on the left with neutral ceiling fill, warm approachable corporate grade, extreme detail, photorealistic. Sticky notes and whiteboard carry no readable writing.

提示词
Two hands clasped in a firm handshake over a light oak conference table, one in a charcoal suit cuff with a plain watch, the other in a navy sleeve, a closed leather folio and two water glasses just visible at the edges of frame. Faces out of frame. Close-up, slightly high angle looking down at the hands, 50mm lens, shallow depth of field. Soft window light from the left with gentle bounce fill, warm neutral corporate grade, extreme detail, photorealistic.
适合场景
适合
- 文字密集的视觉——海报、信息图、图解与含标注的版面
- 拟真的商品与营销摄影图
- 自定义宽高比,例如宽屏横幅与竖版长图
通常较不适合
- 图像编辑或参考图工作流——ATP 目前仅将此 id 列为文生图
- 重视稳定性的 2K 以上输出——OpenAI 将超过 2560×1440 的尺寸视为实验性
每个模型都用同一套管理:项目密钥、各项目可用的模型、点数预算,每次请求都有记录。设置组织
成本试算
拉一下用量,看一个月大概花多少。
010
050
试算
$62.50
USD / 月
6,250 点数 / 月
- 文本输入$2.50
- 图片输出$60.00
$62.50 USD / 月 · 6,250 点数 / 月
快速上手
先确认你的密钥能看到这个模型代号,再按文档发请求。 阅读文档
models.sh · curl
curl -s https://api.atptoken.ai/v1/models \
-H "Authorization: Bearer atp-•••"
# data[].id: "gpt-image-2"与同级模型比较
同类型、同定位、其他厂商的模型,价格最接近的排前面。
常见问题
- GPT Image 2 会生成什么?
- GPT Image 2 是 OpenAI 的图像生成模型。生成图片。可用的输入类型请看图像生成文档。
- GPT Image 2 怎么计价?
- 按每百万文本输入 Token 和每百万图片输出 Token 计价。请求成功才扣点,失败不收费。USD 1=100 点。 查看价格
- 怎么通过 ATP 调用 GPT Image 2?
- 创建项目 API 密钥,先用 GET https://api.atptoken.ai/v1/models 确认能看到「gpt-image-2」,再按图像生成文档发请求,model 填这个代号。
资料来源
本页的模型介绍引用以下来源。价格来自 ATP。