gpt-image-2.6-flare

由 OpenAI 提供图像

GPT Image 2.6 Flare 支持官方渠道同步生图和编辑,可设置画质、尺寸、透明背景及输出格式。

更新于:
输入文本和图像编辑最多使用 16 张参考图
输出4 种格式PNG · JPEG · WebP · GIF
返回方式同步响应直接返回图像
价格优惠 20%官方价
Night-market vendor portrait — GPT Image 2.6 Flare published exampleNight-market vendor portrait示例 01
Illustrated night-market map — GPT Image 2.6 Flare published exampleIllustrated night-market map示例 02
Astronaut cat sticker — GPT Image 2.6 Flare published exampleAstronaut cat sticker示例 03
Dumplings to go — GPT Image 2.6 Flare published exampleDumplings to go示例 04
生成示例

从提示词到成图

Candid photo of a street food vendor in her fifties laughing mid-flip as she turns skewers over a charcoal grill at a crowded night market, smoke rising through strings of bare bulbs. Shot on a phone with direct on-camera flash, real skin texture, slight motion blur, muted true color with one chili red accent, candid and unposed, no text, logos, or watermark.

Close-up of a hand holding an open kraft takeaway box of pan-fried dumplings toward the camera at a night market, a round sticker on the lid reading exactly "STEAMED DAILY". Shot on a phone with direct on-camera flash, crisp grease sheen, slight grain, muted true color with one mustard accent, sharp legible sticker, no other words or logos.

A customer in a yellow raincoat waiting under a clear plastic awning at a night market wok stall in heavy rain, neon reflections on the wet pavement, steam rising from the wok. Shot on a phone with direct on-camera flash, rain streaks caught in the flash, slight grain, muted true color with one yellow accent, candid and unposed, no readable text, logos, or watermark.

Illustrated map poster of a night market on cream paper, titled exactly "NIGHT MARKET", six numbered stalls drawn along two aisles, with a legend reading exactly "1 Noodles", "2 Dumplings", "3 Skewers", "4 Bubble Tea", "5 Bao", "6 Mango Ice". Clean editorial design, muted earth tones with one chili red accent, subtle paper texture, sharp legible type, no other words or logos.

Glowing lightbox menu above a night market noodle stall, reading exactly "HAND-PULLED NOODLES" with three lines below: "Beef Broth 12", "Chili Oil 10", "Wild Mushroom 11". Steam drifting past the sign. Clean editorial design, muted true color with one chili red accent, sharp legible type, no other words or logos.

A clean vertical infographic poster titled "How Coffee Gets to Your Cup". Five numbered steps run top to bottom, each with a small flat illustration and a one-line caption: "1. Harvest" with ripe red cherries on a branch, "2. Process" with beans drying on raised beds, "3. Roast" with a drum roaster, "4. Grind" with a burr grinder, "5. Brew" with a pour-over dripper. Thin arrows connect each step. Warm cream background, dark brown sans-serif type, terracotta accents. Footer text "Farm to cup in 5 steps". No other text.

A die-cut vinyl sticker of a cheerful cartoon astronaut cat floating with a tiny jetpack, bold black outline, thick white sticker border, flat bright colors, soft drop shadow, isolated subject

0.5x ultra-wide phone shot looking down a packed night market aisle from a footbridge, rows of stalls under strings of bare bulbs, steam and crowds, a kid on a parent's shoulders looking up at the camera. Harsh mixed light, slight grain, muted true color with one amber accent, candid and unposed, no readable text, logos, or watermark.

模型能力

gpt-image-2.6-flare 能做什么

文生图

通过所选模型的文档接口,用文字生成照片、插画、示意图和宣传素材。

官方渠道输出设置

官方生成接口的画质、尺寸、背景透明度和格式选项因模型而异。

按渠道处理结果

官方生成接口同步返回图像,固定价渠道则提交异步任务,再查询结果。

按模型构造请求

查看所选模型的请求与操作列表,确认支持的字段和返回方式。

计费

按操作计费,单位为 quota

按模型、操作和输出设置扣除 quota。

官方渠道图像生成与编辑官方定价
模型价格
gpt-image-2.6-flare比官方价低 20%
API 请求

按操作查看请求

选择一个操作,查看对应接口和请求字段。

OpenAI Image

生成

通过官方渠道同步文生图。

POSThttps://api.ttapi.io/v1/images/generations

请求头

TT-API-KEYstring必填

填写 TTAPI API 密钥。

Content-Typestring必填

使用 application/json。

请求体

promptstring必填

图像生成描述,最多 32,000 字符。

modelenum必填

此模型使用 gpt-image-2.6-flare。

backgroundenum选填

可选 transparent、opaque 或 auto,默认 auto。透明背景仅支持 png 或 webp。

moderationenum选填

内容审核强度可设 low 或 auto,默认 auto。

ninteger选填

生成 1 至 10 张图像,默认 1 张。

output_compressioninteger选填

范围 0 至 100,默认 100,仅用于 jpeg 或 webp 输出。

output_formatenum选填

可选 png、jpeg、webp 或 gif,默认 png。仅 GPT Image 2.6 支持 GIF。

partial_imagesinteger按条件填写

流式返回 0 至 3 张中间图像,默认 0。仅在 stream 为 true 时可用。

qualityenum选填

可选 low、medium、high、xhigh、max 或 auto,默认 auto。

response_formatenum选填

仅支持 b64_json。GPT Image 始终返回 Base64 图像数据。

sizestring选填

可选 auto(默认)或 WIDTHxHEIGHT。宽高必须为 16 的倍数,宽高比介于 1:3 至 3:1,最长边不超过 3840 像素,总面积须为 655,360 至 8,294,400 像素。

streamboolean选填

默认 false。开启流式响应可接收生成进度和中间图像。

userstring选填

用于识别滥用的唯一终端用户标识。

查看当前渠道的接入文档
接入示例

查看选中渠道的代码示例

选择渠道和操作,再复制相应的接口地址、请求体或代码示例。

OpenAI Image

生成 接入示例

POST https://api.ttapi.io/v1/images/generations

curl --request POST \
  --url https://api.ttapi.io/v1/images/generations \
  --header "TT-API-KEY: $TTAPI_KEY" \
  --header 'Content-Type: application/json' \
  --data '{"model":"gpt-image-2.6-flare","prompt":"A premium product photograph of a ceramic coffee cup on a warm gray background, soft window light, no text.","size":"1024x1536","quality":"medium","output_format":"webp","n":1}'