Create animated GIFs with Sunburst or Flare, guide an edit with a transparent annotation layer, and choose how much planning goes into each image.
No GPU. No setup. Cancel anytime.
GPT Image 2.6 brings GIF animation to OpenAI image generation. Describe a short action and its loop, choose Sunburst or Flare, and receive an animated GIF. Sunburst targets detailed work; Flare targets faster iteration.
Still-image work adds planning complexity and annotation-guided editing. Send a separate transparent PNG with arrows, circles, or labels over the first reference, then describe what should change. Complexity controls planning separately from rendering quality.
Both run through Sogni on the Fast network and use Premium Spark. Existing GPT Image models remain available. Compare your own prompts: our small Medium-quality sample found faster Flare responses and slower Sunburst responses than their 2.5 counterparts.
Choose either variant explicitly. Both expose the same formats, quality settings, and planning controls.
| Model | Details | From | |
|---|---|---|---|
Sunburst Qualitygpt-image-2.6-sunburst | — | 11.3 Spark · $0.057 | Create → |
Flare Fastgpt-image-2.6-flare | — | 11.3 Spark · $0.057 | Create → |
896 × 896 · Nine frames at 200 ms · Medium quality · High complexity
Read the API guide · Compare GPT Image 2.6 with 2.5: GIFs, price and speed
GPT Image 2.6 is a pay-as-you-go Premium Spark vendor model. Every render uses Premium Spark (bought or awarded, never Free Spark), including for Sogni Unlimited members. Unlimited saves 5%; Unlimited Pro saves 10%.
| Variant / configuration | Spark | USD |
|---|---|---|
| 1024 × 1024 · medium · output tokens | 11.3 Spark | $0.057 |
| 896 × 896 · 9-frame GIF · medium · observed output tokens | 34.2 Spark | $0.17 |
Premium Spark required. Prices below cover output tokens at the stated size and quality; prompt and reference input add cost.
GIFs: 2–16 frames, 20–2000 ms per frame, with frame-size limits that depend on count. Default: nine 896 × 896 frames at 200 ms each. Text-to-image only.
GIF quotes reserve a conservative output budget; final charges decrease to confirmed usage when lower.
Capabilities and pricing checked November 4, 2026. See the developer guide for parameter details.
One Sogni API key reaches every model in the catalog — call GPT Image 2.6 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-2.6-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 24hcurl 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-2.6-sunburst" }
}]
},
"confirm_cost": true
}' Example uses gpt-image-2.6-sunburst. Use any exact model id from the Variants table above. Full reference at docs.sogni.ai.
GPT Image 2.6 is a third-party frontier model. It uses Premium Spark for every render; Unlimited plans discount those pay-as-you-go rates instead of covering them as fair-use renders.
Vendor-backed renders are billed with Spark even when your account has an active Unlimited subscription.
Unlimited saves 5%; Unlimited Pro saves 10%.
Launch GPT Image 2.6 from the Sogni app or call the exact model id through the API.
Use premium vendor models alongside Sogni Supernet image, video, music, and language models.
ImagePremiumPartner GPT Image 2 OpenAI’s image model through the Sogni API — 16 references, superb instructions. from 13.7 Spark · $0.069 
ImageNew BiRefNet Prompt-free background removal. One image in, a clean cut-out or a soft matte out — always at the size you gave it, for half a cent. from 1.00 Spark · $0.005 Choose GPT Image 2.6 Sunburst or Flare and GIF output. Describe the action and how it returns to the starting pose. The default is nine 896 × 896 frames at 200 ms each, a 1.8-second loop. GIF generation accepts no reference images, masks, or annotation layers.
Low, Medium, and High complexity control planning independently of image quality. High is the default. Choose the rendering quality separately when comparing outputs and prices.
Supply a transparent PNG matching the first reference image and describe the edit. The layer guides where the change applies. It is separate from the source photo and cannot be combined with a mask.
The per-token rates are the same, but output token budgets differ. At 1024 × 1024 Medium, our requests returned 439 output tokens for 2.5 and 1,449 for 2.6. Quality, size, prompts, references, and GIF frame count affect the quote.
In three Medium-quality 1024 × 1024 prompts per model, direct-provider median time was 6.55 seconds for 2.6 Flare versus 10.48 for 2.5 Flare; Sunburst was 19.99 versus 14.96 seconds. This small sample excludes Sogni delivery and is not a general performance ranking.
PNG and WebP output are still images and can use transparent backgrounds. Animation output is GIF, with an opaque background. Transparent animated PNG (APNG) generation is not supported.
Create in the app, or build with the API. Your call.