GPT Image 2.5 Sunburst and Flare

About

GPT Image 2.5 is OpenAI’s newest image family, available through the same Sogni key as every other model. It comes in two sizes. Sunburst is OpenAI’s quality-first model, which OpenAI rates above GPT Image 2 for image quality. Flare is the smaller, faster model, with image quality OpenAI rates comparable to GPT Image 2.

Both add three things GPT Image 2 does not have: two higher quality settings, Extra high and Maximum; a transparent background, which returns a cut-out as PNG or WebP; and an edit mask — a PNG with transparency, the same size as your first reference image — that confines the change to the area you mark. Like GPT Image 2, they take up to sixteen reference images and render text well.

As partner models they run on OpenAI’s infrastructure rather than Supernet GPU workers and are billed in Premium Spark only. Sogni quotes each job before it runs, and the final charge follows the output tokens OpenAI reports, so quality, size and references move the price. GPT Image 2 stays available, unchanged.

Two models, one family

Sunburst and Flare take the same inputs and settings, and Sogni quotes the same request at the same price on either. Choose Sunburst for the best result and Flare when speed matters more.

Speed test: GPT Image 2 vs 2.5 Flare vs 2.5 Sunburst

On September 24, 2026 we ran the same five prompts through all three models on Sogni at every quality setting: a portrait with a lettered mug, a chalkboard menu, an alpine lake, an infographic and a watercolor fox. We also gave them the Sogni Photobooth’s sample photo of Einstein to transform. That came to 168 images. The models started each prompt together, so they faced the same conditions. Time runs from submitting the job to the finished image being ready to download.

Median time across the five prompts, then the pay-as-you-go price of one 1024 × 1024 image. Unlimited saves 5% and Unlimited Pro 10%. Low: 10 images per model; Medium and High: 15, with one GPT Image 2 High job lost to a temporary OpenAI error; Extra high and Maximum: 5.

From Medium up, GPT Image 2.5 is two to five times faster. At Low all three finish in 12 to 14 seconds for the same price. At Medium, Flare took 14 seconds and Sunburst 17, against 37 for GPT Image 2; at High, 21, 34 and 105. Flare was the fastest at every setting.

It costs a quarter as much. At Medium and High, GPT Image 2.5 costs a quarter of GPT Image 2’s price, so its High costs what GPT Image 2’s Medium does and still finishes sooner.

Text held up on all three. All 28 chalkboard menus spelled every item and price correctly, and all 27 mugs read SOGNI STUDIO. The one repeated miss was Flare on the infographic: from Low to High it dropped the step numbers in 7 of 8 images, captioning SEED instead of 1 SEED. GPT Image 2 and Sunburst kept them every time.

Below is the first of three takes for every prompt at Medium and High, not a pick of the best. Click any image for the full-resolution file.

Medium (HQ)High (Pro)

Portrait with a lettered mug 1024 × 1024

Photorealistic editorial studio portrait of an adult ceramic artist with curly dark hair and natural skin texture, wearing a navy knitted sweater. She holds a small ivory ceramic mug at chest height with both hands naturally visible. The mug has a single centered label reading exactly SOGNI STUDIO in crisp dark blue uppercase letters. Eye-level medium close-up, face and mug both in sharp focus. Soft diffused studio lighting, realistic glazed ceramic highlights and detailed wool fibers. The background is a smooth blue-to-peach gradient. One person, one mug, balanced square composition.

Chalkboard menu: six items, six prices 1024 × 1536

Photograph of a café chalkboard menu hanging on a red brick wall, straight-on view, soft window light from the left. The heading at the top reads exactly MORNING MENU in large hand-lettered chalk capitals. Below it are six lines of neat white chalk lettering, each item on the left and its price aligned on the right: Espresso 3.00, Flat White 4.50, Oat Latte 5.25, Cold Brew 4.75, Almond Croissant 4.25, Blueberry Scone 3.75. A small chalk drawing of a steaming coffee cup sits in the lower left corner and a croissant in the lower right corner. Visible chalk dust, faint smudges from earlier erasing, and a natural wooden frame.

Landscape detail 1536 × 1024

Wide landscape photograph at dawn. A perfectly still alpine lake reflects jagged granite peaks with patches of snow. In the foreground a weathered wooden rowboat is tied to a small dock with frayed rope; the peeling paint, wood grain and iron nails are sharp and detailed. Dark pine forest lines the shore, a thin layer of mist drifts over the water, and the sky is a smooth gradient from deep blue at the top to pale apricot at the horizon. Natural colors, sharp from the foreground to the distant peaks.

Infographic with exact captions 1024 × 1024

A clean flat vector infographic on a cream background. The title at the top reads exactly HOW A SEED BECOMES A TREE in bold dark green sans-serif capitals. Below it, four panels in a row connected by arrows, each with a simple illustration and a caption beneath it: 1 SEED shows a brown seed in soil, 2 SPROUT shows a small green shoot with two leaves, 3 SAPLING shows a thin young tree, 4 TREE shows a full leafy oak. At the bottom, one line of smaller text reads exactly Water, sunlight and time. Consistent line weights, a limited palette of greens and browns, generous spacing.

Watercolor illustration 1024 × 1024

A whimsical watercolor illustration of a red fox sitting under a large red-capped mushroom during a light rain, reading a small open book held in both paws. Loose, transparent watercolor washes with soft blooms and granulation, fine ink linework on the fox fur and whiskers, visible cold-press paper texture, raindrops drawn as short pale strokes, puddles reflecting the mushroom, muted greens and warm oranges, plenty of white paper around the scene.

Portrait edits. We reproduced a Sogni Photobooth render: its sample photo of Einstein, its Viking and claymation styles and the prompt the app builds, at 1024 × 1536. At Low, the Photobooth’s default quality, GPT Image 2.5 Flare took 14 seconds and cost $0.018 an image, against 18 seconds and $0.047 on GPT Image 2. All three kept Einstein recognizable; GPT Image 2 kept the photo black and white, and both 2.5 models colorized it.

Photobooth Viking edit at Low 1024 × 1536

Measured through Sogni’s standard image API on the fast network: one image per job, PNG output for the five prompts, JPEG with one reference image for the portrait edits. Times include Sogni’s queue, OpenAI’s render and delivery. A few takes per setting show a typical time, not a guarantee.

Not tested in this section: edit masks and larger sizes (both are in the article), transparent backgrounds and several reference images.

Every setting, all 168 images, the full Photobooth comparison and our edit and mask tests at 2688 × 1536 are in the article: GPT Image 2.5 vs GPT Image 2: Speed, Price and Quality Tested.

Pricing

GPT Image 2.5 is a pay-as-you-go Premium Spark vendor model. Every render uses purchased Spark, including for Sogni Unlimited members. Unlimited saves 5%; Unlimited Pro saves 10%.

Premium Spark required — purchased credits; free Spark does not apply.

In the app, low, medium and high are called Fast, HQ and Pro; GPT Image 2.5 adds Extra high (xhigh) and Maximum (max).

Pricing is token-based: the quote comes first, and the charge follows the output tokens OpenAI reports. Reference images add input-token cost.

API

One Sogni API key reaches every model in the catalog — call GPT Image 2.5 with the exact model id for the variant you want.

import { SogniClient } from '@sogni-ai/sogni-client';

const client = await SogniClient.createInstance({
  appId: crypto.randomUUID(),
  apiKey: process.env.SOGNI_API_KEY,
  network: 'fast',
});

const project = await client.projects.create({
  type: 'image',
  modelId: 'gpt-image-2.5-sunburst',
  positivePrompt: 'a slothicorn surfing a wave of liquid paint, studio lighting',
  numberOfMedia: 1,
});

const [url] = await project.waitForCompletion();
console.log(url); // result link — download within 24h

curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -d '{
    "input": {
      "title": "First render",
      "steps": [{
        "id": "step1",
        "toolName": "generate_image",
        "arguments": { "prompt": "a slothicorn surfing a wave of liquid paint, studio lighting", "model": "gpt-image-2.5-sunburst" }
      }]
    },
    "confirm_cost": true
  }'

Example uses gpt-image-2.5-sunburst. Use any exact model id from the Variants table above. Full reference at docs.sogni.ai.