HiAPI
21:9 电影宽幅 Hero:左侧柔光留白,右侧悬浮多层虹彩玻璃图像板与陶瓷光环,gpt-image-2.5 为发光立体字,冷蓝粉金多色渐变。作为 gpt-image-2.5-flare 详情页顶部横幅。

OpenAI

GPT Image 2.5

gpt-image-2.5-flare/text-to-image

OpenAI's two-tier image generation and editing models: Flare focuses on speed, Sunburst pursues precision, covering both text-to-image and multi-turn image refinement

Route

Try gpt-image-2.5-flare/text-to-image

Input

Output

image
Fill in the parameters and run — results show up here.

Pricing

gpt-image-2.5-flare/text-to-image per image

Specifications Unit Price
1K $0.050
2K $0.080
4K $0.120

Overview

gpt-image-2.5-flare is the text-to-image line in the OpenAI image generation family, balancing speed and quality, and suited to applications that need to produce images in bulk. It accepts prompts up to 20,000 characters long, allows specifying 1K to 4K resolutions and a variety of aspect ratios, and supports transparent background output at 1K. Compared with the previous generation, this version reduces per-image latency while maintaining higher generation quality, making it suitable for scenarios with throughput requirements such as e-commerce imagery, social media assets, and rapid prototype validation.

Key Highlights

  • A text-to-image line that balances speed and quality — Flare is positioned as the default choice for most applications, delivering noticeably shorter generation times per run while outperforming the previous generation in quality, making it suitable for high-concurrency image generation.
  • Transparent background output directly at 1K — At 1K resolution with background set to transparent, you can directly obtain an image with a transparent background, reducing post-processing steps for cutout and compositing.
  • Flexible resolution and canvas options — Supports three resolution tiers (1K, 2K, 4K) and more than ten aspect ratios, so you can directly select the image specification according to the delivery channel.

Capabilities

Long-prompt text-to-image: A single prompt can be up to 20,000 characters, making it easy to describe complex layouts and structured information; images are produced directly from text without the need to prepare reference images.

Resolution and aspect ratio control: You can choose an output tier among 1K, 2K, and 4K, and specify the canvas from ratios such as 1:1, 16:9, 21:9, and 9:16, generating the appropriate specification in one pass for the target placement.

Transparent background output: At 1K resolution, you can set background to transparent to generate an image with a transparent background, suitable for asset overlays and downstream compositing workflows.

Use Cases

  • E-commerce and operations teams generating product imagery in bulk — Using product selling-point text as input, produce images in bulk at channel-specific ratios for direct use in product detail pages and ad assets.
  • Social media creators quickly producing content images — Use short prompts with 9:16 or 1:1 ratios to generate accompanying images, shortening the cycle from topic selection to publishing.
  • Product teams validating visual prototypes — Before requirements reviews, use low-cost, low-latency text-to-image to quickly compare multiple visual directions, then move to refinement after the direction is decided.
  • Designers generating transparent-background assets — Use the 1K transparent background mode to produce graphic elements that can be overlaid directly for posters, pages, and video compositing.

Limitations and Selection Advice

This line supports text-to-image only; it cannot edit based on reference images or perform repeated refinement, so such needs should switch to the image editing line in the same family. Transparent background is available only at 1K; when 2K or 4K is selected, the background mode cannot be specified. 4K output incurs higher computation and time overhead, so for low-latency batch scenarios, 1K or 2K is recommended first. Layouts containing dense small-font text or complex legends remain error-prone scenarios and require human review before delivery; the model is a closed-source API and does not support local deployment or fine-tuning, and content usage must comply with the corresponding compliance requirements.

Related Models

View all models →

Same Type