NezhaGateNezhaGate
AnnouncementsAll updates →
NoticeFrom October 29: new top-up bonuses ($50 gives 300 credits) and a $1.40 refund fee NoticeWan 3.0: Standard 480P / 720P / 1080P available; Prime paused NewKling 4.0 is live: reference images, videos and audio, 3-30 seconds, billed per second at 720P / 1080P
简体中文English ✓Español한국어日本語Русский
Log in Get API Key
Image ● Operational Service status →

GPT Image 2.5 Sunburst

⧉
gpt-image-2.5-sunburst
24H STATUS Success: 99.7%
-24hnow
Real customer calls only, counting successes and our own failures: green is success in that hour, red is the failure share. Three things are excluded -- client-side errors (bad request, auth, insufficient balance and other 4xx), requests still in flight, and our own internal test calls. One bar per hour, taken from real gateway call logs.
Model Type FlareSunburst

GPT Image 2.5 Sunburst is the other style branch of the next-generation OpenAI image model. Against Flare it comes back with heavier grain and brushwork and firmer highlights, carrying more visual information - a fit when you want material and physical presence. Capabilities and price are identical to Flare. Text-to-image and image-to-image, 1K / 2K / 4K, with a C2PA content credential in every render. Set model to gpt-image-2.5-sunburst. Billed per image, failed jobs refunded in full.

Text-to-imageImage-to-image1K/2K/4KTexturedC2PA credential
Get API Key → API Docs

Live Test · Playground

Try out GPT Image 2.5 Sunburst right here (available after login).

Model Type

Input

Image

Output

image
Results will appear here after running.
🕑 Generated images and videos are kept for 15 days before 2026-10-29 09:49:40 UTC; from that time, they are kept for 7 days and existing files older than 7 days are removed.

About GPT Image 2.5 Sunburst

GPT Image 2.5 Sunburst is one of two style branches of the next-generation OpenAI image model; its output is grainier, with heavier brushwork and firmer highlights. NezhaGate serves it through the OpenAI-compatible images API: /v1/images/generations for text-to-image and /v1/images/edits for image-to-image, at 1K / 2K / 4K and multiple aspect ratios, with a C2PA content credential embedded in every render. Set model to gpt-image-2.5-sunburst. Billed per image, failed jobs refunded in full, at exactly the same price as GPT Image 2.

Use cases

Covers and posters

Article covers, event posters and product shots, with accurate typography; 1K is already enough for web and social.

Image-to-image revisions

Put an existing image in the image parameter or use /images/edits to repaint a region, swap the background or shift the style.

Batch generation

Async jobs plus direct links (kept for 15 days) make scripted bulk production easy, and failures cost nothing.

When provenance matters

Every render carries a C2PA signature, so you can state that the asset is AI-generated and keep an audit trail.

How to choose

Flare and Sunburst cost the same and do the same things; only the look differs. Pick Flare for clean layouts and legible type, Sunburst for grain and material feel. If you just need ordinary output, the existing gpt-image-2 is priced the same and remains a fine choice.

FAQ

Flare or Sunburst?
Capabilities, resolutions and price are identical; only the look differs. Flare is cleaner and smoother, which suits covers and posters that need a crisp layout and legible type. Sunburst has heavier grain and brushwork with firmer highlights, for images that want material presence. The fastest way to choose is to run the same prompt through both.
Which resolutions and ratios are supported?
1K / 2K / 4K, and size accepts either an aspect ratio (1:1, 3:4, 2:3, 9:16, 4:3, 3:2, 16:9) or a pixel size. The image is delivered at exactly the size you choose: with a ratio, the tier sets the longest side to 1024 / 2048 / 4096; with a pixel size such as 1024x1024, you get that size (longest side up to 4096).
How do I call it?
As an async job: POST /v1/images/generations (text-to-image) or /v1/images/edits (image-to-image) returns HTTP 202 with a job id right away; poll GET /v1/images/jobs/{id} until status=succeeded and read data[0].url, a re-hosted link kept for 15 days (download it before then). Do not read data straight after submitting.
Am I charged for failed jobs?
No. Only jobs that actually return an image are settled; upstream errors and timeouts are refunded in full and never touch your balance.

Related Models

Explore other models you can integrate.

View all →
GPT Image 2.5 Flare Image

GPT Image 2.5 Flare

gpt-image-2.5-flare

Next-gen image model - clean and smooth

$0.005/img and up View →
GPT Image 2 Image

GPT Image 2

gpt-image-2

High-quality text-to-image / image-to-image model

$0.005/img and up View →