GPT Image 2.6 변형과 파라미터
| 항목 | GPT Image 2.6 Sunburst | GPT Image 2.6 Flare |
|---|---|---|
| 길이 | — | — |
| 화면비 | 1:1 · 16:9 · 9:16 · 4:3 · 3:4 · 3:2 · 2:3 · 4:5 · 5:4 · 21:9 | 1:1 · 16:9 · 9:16 · 4:3 · 3:4 · 3:2 · 2:3 · 4:5 · 5:4 · 21:9 |
| 해상도 | — | — |
| 네이티브 오디오 | — | — |
| 이미지→영상 | — | — |
| 참조→영상 | — | — |
| 초당 크레딧 | — | — |
| 5초 클립 비용 | 2 cr/image | 2 cr/image |
GPT Image 2.6은(는) 무엇인가요?
GPT Image 2.6 is the newest generation of OpenAI's image model, released in November 2026, and FlyAIgh offers both of its lines. Like the generation before it, it is the model to reach for when readable text is part of the picture — posters, ad creatives, menus, product labels, slides and infographics. It draws clean letterforms where most image models produce lookalike glyphs, and in our own tests it set a bilingual poster — Chinese headline, English subtitle and a three-row schedule table — with every character as written.
It ships in two lines that cost exactly the same. **Sunburst** is the precision line: it holds intricate detail across multi-round edits and is the pick for finished campaign visuals and product imagery. **Flare** is the faster line for when response time matters. Both take five quality tiers (Low, Medium, High, Extra high, Max), render at 1K, 2K or 4K (4K is 2880 × 2880 for a square), support ten aspect ratios from 1:1 to 21:9, and do both text-to-image and editing with up to 16 reference images.
What is new in 2.6 is a **Complexity** setting — Low, Medium or High, with High as the default — that decides how much the model plans before it draws. High helps with dense text and complex layouts; Low is quicker and suits focused edits such as swapping a background. It costs nothing extra: we generated the same image at Low and High and were charged the same. A transparent background is available when you edit a reference image, at Medium quality or above; the supplier does not render alpha from a text prompt alone.
The price depends on quality and resolution only — aspect ratio and Complexity never change it. Low at 1K is 2 credits, Medium at 1K is 3, High at 2K is 6, Extra high at 4K is 17 and Max at 4K is 38, and each reference image adds about 0.4 credit. The numbers follow OpenAI's published token formula at the supplier's per-token rates, and we checked them against real charges before launch. The exact total is shown before you generate.
Compared with GPT Image 2.5 at square sizes, 2.6 costs more at Low and Medium quality and about the same from High up (High at 1K is one credit more; 2K and above are equal or one to two credits less). 2.5 charges less for wide and tall images and 2.6 does not, so on a 16:9 image 2.5 is the cheaper of the two at every tier. 2.5 also keeps five extra extreme aspect ratios (down to 1:3) and transparent text-to-image, and it has its own page alongside GPT Image 2. Like every model on FlyAIgh, 2.6 runs on the same credit wallet — no OpenAI subscription or API key.
GPT Image 2.6 vs GPT Image vs Nano Banana
| 기능 | GPT Image 2.6 | GPT Image | Nano Banana |
|---|---|---|---|
| 네이티브 오디오 | |||
| 이미지→영상 | |||
| 참조→영상 | |||
| 첫/마지막 프레임 |