gpt-image-2.5-sunburstgpt-image-2.5-sunburst is the tier for images somebody is going to look at closely. It shares the whole GPT Image 2.5 feature set with Flare — point at a region and only that region changes, the same character or product holds together over a long chain of edits, native transparent PNG, native 1K / 2K / 4K in six aspect ratios — and spends roughly twice as long per image to hold the detail Flare only approximates: skin and fabric texture, dense typography, machined and reflective surfaces, and anything heading for print or a full-bleed product page. Quality defaults to high. It is also the tier to reach for when one image goes through many rounds of edits and every round has to leave the rest of the frame untouched. Worth knowing before you pick a tier: Sunburst and Flare cost exactly the same at a given resolution, and quality does not change the price either — any apparent gap between the two pages comes only from their default quality. You are paying for time, not a different rate card. Sketch input, the creation templates and comment-on-image editing are ChatGPT features at launch and are not part of the API.
Generated image will appear here
Enter a prompt and click Generate
Priced on two axes, resolution and quality — fifteen cells. 1K: $0.008 / $0.025 / $0.045 / $0.08 / $0.18. 2K: $0.012 / $0.028 / $0.09 / $0.16 / $0.35. 4K: $0.02 / $0.045 / $0.15 / $0.26 / $0.58 (low to max). You pay for the detail you ask for. The tier follows the longest edge of the image you get back, not the parameter you sent. Transparent background, mask and reference images are included; failed requests are not charged.
Mark the part you want changed and the rest of the frame is left alone — swap an outfit without the face shifting, rewrite on-image text without the background moving. This is the change most people notice first.
The same character or product holds together over a long chain of edits instead of drifting after a few. Keeping a set on-model no longer depends on structural prompts, reference images and fixed-lighting clauses.
Real alpha, not a flat colour you have to key out. GPT Image 2 could not do this, so a transparent-background request is the simplest way to tell the two models apart.
The two tiers cost exactly the same at every size, and now that quality no longer affects price there is no pricing angle to weigh at all. Sunburst simply spends longer per image; pick it when the result will be looked at closely.
Skin pores and fabric weave, dense small type, brushed metal and reflections — the places where Flare rounds detail off and Sunburst does not. If the image ends up on a full-bleed product page or in print, this is the tier.
Sketch (draw a rough layout and let the model read it), the 15 creation templates and commenting directly on an image are ChatGPT features at launch. They are not exposed through the API — worth knowing before you build against them.
GPT Image 2.5 Sunburst is a Image Generation API provided by OpenAI. gpt-image-2.5-sunburst is the tier for images somebody is going to look at closely. It shares the whole GPT Image 2.5 feature set with Flare — point at a region and only that region changes, the same character or product holds together over a long chain of edits, native transparent PNG, native 1K / 2K / 4K in six aspect ratios — and spends roughly twice as long per image to hold the detail Flare only approximates: skin and fabric texture, dense typography, machined and reflective surfaces, and anything heading for print or a full-bleed product page. Quality defaults to high. It is also the tier to reach for when one image goes through many rounds of edits and every round has to leave the rest of the frame untouched. Worth knowing before you pick a tier: Sunburst and Flare cost exactly the same at a given resolution, and quality does not change the price either — any apparent gap between the two pages comes only from their default quality. You are paying for time, not a different rate card. Sketch input, the creation templates and comment-on-image editing are ChatGPT features at launch and are not part of the API. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: 1K-low: $0.008, 1K-medium: $0.025, 1K-high: $0.045, 1K-xhigh: $0.08, 1K-max: $0.18, 2K-low: $0.012, 2K-medium: $0.028, 2K-high: $0.09, 2K-xhigh: $0.16, 2K-max: $0.35, 4K-low: $0.02, 4K-medium: $0.045, 4K-high: $0.15, 4K-xhigh: $0.26, 4K-max: $0.58.
Watch dials, stitching, brushed metal, glass and liquid — the surfaces where Flare rounds detail off and a buyer zooms in anyway. The extra time Sunburst spends goes exactly here.
Ingredient lists, legal small print and dense label typography stay legible at 4K instead of dissolving into texture. That is the difference between a mockup and a file you can hand to a printer.
One image, a dozen art-direction notes, and every round changes only what was asked. Over a long chain of edits Sunburst holds the rest of the frame tighter than Flare does.
Skin that keeps pores and stray hairs instead of going waxy, fabric that keeps its weave. When the final crop is a face at full width, this is the tier.
Native 4K (3840×2160) that still holds at arm's length on a poster or a trade-show panel. $0.02 to $0.58 per image at 4K depending on the quality tier — the same rates Flare pays, so between the two models you are buying time, not resolution
GPT Image 2.5 Sunburst is available through APIMODELS at: 1K-low: $0.008, 1K-medium: $0.025, 1K-high: $0.045, 1K-xhigh: $0.08, 1K-max: $0.18, 2K-low: $0.012, 2K-medium: $0.028, 2K-high: $0.09, 2K-xhigh: $0.16, 2K-max: $0.35, 4K-low: $0.02, 4K-medium: $0.045, 4K-high: $0.15, 4K-xhigh: $0.26, 4K-max: $0.58. Billing is pay-as-you-go — you only pay for what you generate.
Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.
APIMODELS offers the same GPT Image 2.5 Sunburst model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
Sunburst is the highest-fidelity tier of 2.5: finer detail, better materials and closer prompt adherence, rendering at high by default and taking longer per image (about 50s at 1K, 100s at 2K). It is built for premium commercial visuals and print work — images that have to survive being looked at up close. Flare is the default tier: medium by default, about half the latency of GPT Image 2. Parameters and pricing are identical, so if your requests already pass quality:"high" the remaining difference is the upstream fidelity tuning.
Priced on two axes — resolution AND quality — fifteen cells in all. 1K: $0.008 / $0.025 / $0.045 / $0.08 / $0.18. 2K: $0.012 / $0.028 / $0.09 / $0.16 / $0.35. 4K: $0.02 / $0.045 / $0.15 / $0.26 / $0.58 (low / medium / high / xhigh / max). You pay for the detail you actually ask for: a layout check at 1K low is $0.008, and you only reach for max when the image is going to print. The tier follows the longest edge of the image you actually get back, not the parameter you sent (with size auto or omitted, the output decides). Reference images, mask and transparent background cost nothing extra, and failed requests are not charged. With quality omitted or auto you get the model's default — medium for flare, high for sunburst — and both tiers cost the same, so you are choosing speed, not price.
It depends on the tier you use. OpenAI bills 2.5 per output token ($30 per million image tokens), so higher quality costs more — the same 4K image is about $0.011 at low and about $0.40 at max, a 36x spread. We bill on the same two axes, so every cell compares directly: 4K max is about $0.40 at OpenAI versus $0.58 here; 2K high about $0.107 versus $0.09; 1K low about $0.006 versus $0.008. Same order of magnitude overall — we are cheaper through the middle and a little dearer at the very top, where the upstream cost has to be covered. If you run max or xhigh, or simply do not want to re-cost every quality decision, this is the better deal. You also skip OpenAI organisation verification, and one key covers every model on the site.
Transparent backgrounds do: background=transparent with output_format=png returns a real alpha PNG, verified at 1K, 2K and 4K (52-67% transparent pixels on product shots). A mask is accepted (the mask field of multipart /v1/images/edits, a PNG the same size as the image), but in our tests 2.5 treats it as a hint rather than a hard boundary: the prompt drives the edit scope. Name the object to change and say the rest must stay untouched and local edits are reliable; a mask alone will not confine the change precisely.
Yes. 2048×2048 and 3840×2160 are delivered exactly as requested, and we checked them with a detail-energy test (downscale by half, upscale back, measure what is lost): the 2K and 4K outputs carry as much full-resolution high-frequency detail as a native 1K image, so they are not upscales. As a side note, token-based billing makes 4K slightly cheaper than 2K — that is OpenAI's accounting, not an error.
It scales with quality: low about 30-45s, medium 40-70s, high about 50s at 1K, 100s at 2K and 70s at 4K; reference-image edits 15-50s. For synchronous calls set client timeouts to 180s or more, or use the async taskId polling / webhook contract instead.
On APIMODELS, GPT Image 2.5 Sunburst runs alongside 60+ models on one API key and one balance, so choosing is about fit, not lock-in. It supports Text to Image, Precise region edit — mark a spot, only that changes, Multi-turn consistency without prompt scaffolding, Native transparent PNG, Sharper detail, faster generation, 1:1 / 3:4 / 4:3 / 9:16 / 16:9 / 21:9, Flare (default, ~2x faster) and Sunburst (highest fidelity), and you can weigh it on price and capability against other Image Generation models, then switch by changing a single model-name string — no new account or integration. Browse every Image Generation option with live pricing at apimodels.app/models.
GPT Image 2.5 Sunburst supports: Text to Image, Precise region edit — mark a spot, only that changes, Multi-turn consistency without prompt scaffolding, Native transparent PNG, Sharper detail, faster generation, 1:1 / 3:4 / 4:3 / 9:16 / 16:9 / 21:9, Flare (default, ~2x faster) and Sunburst (highest fidelity). See the APIMODELS docs for full parameters and call examples.
Yes. APIMODELS exposes GPT Image 2.5 Sunburst through a single unified API and one key — no separate provider accounts, and no need to handle each provider's regional network access yourself.
We support Stripe (Visa, Mastercard, and other international cards) and Alipay. Credits are available instantly after payment.
Prompts shared by their authors — copy and adapt them. Each one credits its author and links back to the original post.
Weathered sailor on a fishing boat
Create a photorealistic candid photograph of an elderly sailor standing on a small fishing boat. He has weathered skin with visible wrinkles, pores, and sun texture, and a few faded traditional sailor tattoos on his arms. He is calmly adjusting a net while his dog sits nearby on the deck. Shot like a 35mm film photograph, medium close-up at eye level, using a 50mm lens. Soft coastal daylight, shallow depth of field, subtle film grain, natural color balance. The image should feel honest and unposed, with real skin texture, worn materials, and everyday detail. No glamorization, no heavy retouching.
by OpenAI
Automatic coffee machine workflow infographic
Create a detailed Infographic of the functioning and flow of an automatic coffee machine like a Jura. From bean basket, to grinding, to scale, water tank, boiler, etc. I'd like to understand technically and visually the flow.
by OpenAI
Thread streetwear ad with exact typography
Give me a cool in culture ad / fashion shot for a brand called Thread. It's a hip young street brand. The ad shows a group of friends hanging out together with the tagline "Yours to Create." Make it feel like a polished campaign image for a youth streetwear audience: stylish, contemporary, energetic, and tasteful. Use clean composition, strong color direction, natural poses, and premium fashion photography cues. Render the tagline exactly once, clearly and legibly, integrated into the ad layout. No extra text, no watermarks, no unrelated logos.
by OpenAI
We curate copy-ready prompt libraries — every entry shows its full text and a sample result, ready to adapt.
How to get access, regional availability, and how this model compares with its alternatives.