GPT Image 2.6 Generator and Editor

OpenAI released this model on November 4, 2026, and in ChatGPT it is called ChatGPT Images 2.6. This page runs Sunburst for new images and edits. Write a prompt for a still, a looping GIF, or an SVG, or upload up to four references and name one change. There is no zero-credit run. A Low-quality 1024×1024 still costs 3 credits.

Image Editor

(Not effective? Try switching the model and try again!)

Reference Image(0/4)
0 / 4000
Cost 3 creditsRemaining 0 credits

What Is GPT Image 2.6?

Images 2.6 is the release OpenAI says is about 50% faster than the previous version, with steadier characters, products, and brand details across edits. Flare is the speed-first draft model. gpt-image-2.6-sunburst is the higher-detail model this page calls for text to image and editing. gpt-image-2.6-flare is not offered here. Saved reference libraries and frame-by-frame GIF editing stay inside ChatGPT, so each run uses only the files you upload. The plural search gpt images 2.6 and the model id gpt-image-2.6 name the same release.

Sunburst Detail at Every Quality Tier

Sunburst is the higher-detail build. Fabric, labels, skin, and reflections hold up when you zoom in. Five tiers, from Low to Max, decide how much detail a draft is worth. Low judges a composition. Max is for hero shots and large prints.

Edits That Leave the Rest Alone

Name one change, such as a jacket color or a new background, and the framing, face, and lighting stay put. Sunburst is built for consistency across rounds, so a later revision can still match the approved shot. Say which details must not move.

Cleaner Layouts for Posters and Products

Headlines, packaging panels, and product grids come out with more orderly spacing than earlier releases. Put exact wording in quotes, say where it sits, and proof the small print, because tiny lettering can still slip.

Reasoning Effort You Can Set

Reasoning effort controls how much planning happens before rendering. OpenAI recommends Low for most edits, and Medium or High for dense scenes or prompts with many constraints. Higher effort takes longer and does not change the credit cost.

Use Cases for Ads, Products, and Design

Sunburst pays off most on work that goes through review: a brief becomes drafts, and each note should cost one edit. Teams usually explore at Low quality, then save High or Max for the final pass. The panel shows the credit cost before you generate. Searches such as "how can i use chatgpt images 2.6" want that browser workflow. The singular search chatgpt image 2.6 points to this same model.

Generate several hero directions, refine the winner in place, then turn it into a looping GIF for stories, email headers, or display ads. Ask for one clear motion and a still camera. Keep a PNG still for placements that cannot play motion.

GPT Image 2.6 campaign ad of a model holding a water bottle

Settings, Formats, and Limits

Every option below sits in the panel, and the button shows the credit cost before you run. Price follows output size, quality, image count, GIF frames, and reference uploads, because Sunburst is billed by tokens. SVG exists only in text to image, a transparent background needs PNG or WebP, and a GIF shares one pixel budget. Greyed-out options are combinations the current format or size cannot support.

Looping GIF Animation

Pick the GPT Image 2.6 GIF format, then set 2–16 frames and a per-frame duration from 20–2,000 ms, in steps of 10 ms. Frames share an 8,294,400-pixel budget, so 1024×768 allows 10 frames and 1024×1024 allows 7. The largest sizes are unavailable. Each run returns one file.

Up to Four Reference Images

The edit tab accepts up to four references, such as a product, a face, a style board, and a logo. Output size snaps to the closest supported ratio, and each reference adds about one or two credits. Name each file's role in the prompt.

Five Quality Tiers and Reasoning Effort

Quality runs Low, Medium, High, Extra high, and Max. Reasoning effort runs Low, Medium, and High. Quality drives detail and cost. Reasoning drives planning time and leaves the price unchanged, so raise it only for crowded or rule-heavy scenes.

Transparent or Opaque Background

Auto lets the model decide, Opaque returns a finished frame, and Transparent returns a cutout with a real alpha channel. Transparency works with PNG and WebP only. JPEG would flatten the cutout onto a solid fill.

GPT Image 2.6 SVG Export

Text to image can return an SVG file instead of pixels, which suits icons, badges, stickers, and simple logos. Describe flat shapes and a short color palette, since vectors work best for clean graphics rather than photographic scenes. Each SVG run returns a single file.

Sizes Up to 3840×2160

Choose 1:1, 4:3, 3:4, 3:2, 2:3, 16:9, or 9:16. Squares go up to 2880×2880, and widescreen goes up to 3840×2160. Generate up to four stills in PNG, JPEG, or WebP, with 0–100 compression on JPEG and WebP.

GPT Image 2.6 FAQ

Answers on versions, pricing, GIF and SVG output, references, precise edits, and how this page compares with ChatGPT Images 2.6, plus a note on the cheapest way to test a prompt.











Looking for prompt ideas and model comparisons? Read the model guides blog.