Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
As of September 29, 2026, ChatGPT’s user-facing image generator is called ChatGPT Images 2.5. OpenAI presents it as the current image-creation and editing feature in ChatGPT. It is not accurate to say that every ChatGPT image request uses one publicly identified API model ID.
OpenAI’s developer platform uses separate names, including GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst. Those are API model identifiers, while ChatGPT Images 2.5 is the product name shown to users.
The short answer: ChatGPT Images 2.5
When you ask ChatGPT to create or edit a picture, the current feature is branded ChatGPT Images 2.5. OpenAI announced in September 2026 that Images 2.5 is available to ChatGPT, ChatGPT Work and Codex users on desktop, mobile and web.
The name describes the ChatGPT feature, not necessarily a single model ID that developers can select in the API. OpenAI’s public documentation does not establish that every image request made inside ChatGPT is routed through one fixed API identifier. The safest answer is therefore:
#1 Best Overall
- ChatGPT product: ChatGPT Images 2.5.
- Developer API: GPT-Image model identifiers, including GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.
- Historical ChatGPT option: the official DALL·E GPT, which OpenAI says was retired on August 30, 2026.
Is ChatGPT still using DALL·E?
Not as the current named ChatGPT image experience. OpenAI’s Help Center says the official DALL·E GPT in ChatGPT was retired on August 30, 2026, and directs users to ChatGPT Images instead.
DALL·E remains an important historical name because earlier ChatGPT image workflows used it and many tutorials still mention it. Those tutorials can be outdated even when the basic instruction—describe an image, request an edit, or upload a reference—still works. For current ChatGPT use, look for the Images feature rather than trying to open the retired DALL·E GPT.
ChatGPT Images 2.5 versus GPT-Image API models
These names operate at different layers. Confusing them can lead to incorrect assumptions about billing, capabilities or reproducibility.
| Question | ChatGPT Images 2.5 | GPT-Image-2.5 Flare / Sunburst |
|---|---|---|
| Where is it used? | Inside ChatGPT, ChatGPT Work and Codex interfaces | OpenAI’s developer API |
| What is the name? | ChatGPT Images 2.5 | API model identifiers such as GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst |
| Can the public sources establish one fixed ID for every request? | No | These are named API options; availability and behavior depend on the API documentation and account access |
| Is it the retired DALL·E GPT? | No; Images is the current feature | No; these are GPT-Image API names, not the retired ChatGPT DALL·E GPT |
OpenAI may change routing, versions and access without changing the wording users see in ChatGPT. If you need a reproducible developer workflow, select and record the exact API model identifier in your own request rather than assuming the ChatGPT interface exposes the same choice.
What can you do with ChatGPT Images 2.5?
Create images from text
You can describe a subject, composition, style, lighting and intended use in ordinary language. Add constraints that matter—aspect ratio, background color, number of objects, visible text and brand colors—then refine the result conversationally.
Edit an uploaded image
ChatGPT’s image workflow also supports editing. Upload an image, identify the area or object to change, and specify what must remain untouched. For reliable edits, state both the modification and the invariants: for example, “replace the background with a plain light-gray wall; keep the person’s pose, clothing and facial expression unchanged.”
Iterate instead of restarting
Follow-up instructions are useful for targeted changes such as cropping, removing an object, adjusting contrast or changing the visual style. Keeping the conversation together gives the system context about the earlier image and your intended revision.
Exact controls, limits and modes can vary by plan and rollout. OpenAI’s Help Center describes ChatGPT Images as available across all tiers and on web, iOS and Android, while access to related modes may vary by plan and staged availability. If a control is missing, check the account and platform you are using rather than assuming the model has been removed.
Where is ChatGPT Images 2.5 available?
- Platforms: desktop, mobile and web, according to OpenAI’s September 2026 announcement.
- Products: ChatGPT, ChatGPT Work and Codex.
- Plans: Help Center material says ChatGPT Images is available on all tiers, but particular related modes can depend on plan and rollout.
- Regional and account differences: OpenAI can stage features, so two accounts or devices may not show identical controls at the same time.
Availability is a moving product detail. For a current account-specific answer, use the Images entry in your ChatGPT interface and the latest Help Center documentation.
How to identify what generated an image
- Check the interface. If the image was created in the current ChatGPT Images workflow, the user-facing product name is ChatGPT Images 2.5.
- Check your API request. Developer applications should log the model field and response metadata. A request using GPT-Image-2.5 Flare or GPT-Image-2.5 Sunburst identifies the API model selected by that application.
- Do not infer an API ID from a screenshot. An image downloaded from ChatGPT does not, by itself, prove which internal model route produced it.
- Record the date and product. Image systems change. Save the ChatGPT mode or API model name, prompt, input image and relevant settings when reproducibility matters.
Why old articles give different answers
Search results often mention DALL·E because it was the earlier public label, and some developer posts discuss older GPT-Image releases. OpenAI’s December 2025 GPT-Image-1.5 announcement is historical rather than the current answer to this question. Product branding and API identifiers do not necessarily change at the same time.
Another source of confusion is integration news. OpenAI announced in April 2025 that Adobe Firefly and Express would provide access to OpenAI image-generation capabilities. That announcement does not, by itself, establish current access, plan terms or the model used by every Adobe workflow.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Usage scale and what it does not prove
OpenAI reported more than 3 billion images per week in 2026, combining ChatGPT Images activity with GPT-Image models in the API. This is not a ChatGPT-only usage figure, and it does not identify which model handled any individual request. Treat it as a combined OpenAI image-generation statistic, not evidence of a particular routing policy or quality level.
Practical prompt guidance for the current feature
Specify the deliverable
Say whether you need a square icon, a landscape banner, a product mockup, a transparent subject cutout or an illustration for print. Include the intended audience and where the image will appear.
Control text carefully
If the image must contain lettering, provide the exact wording, capitalization and approximate placement. Ask for a clean version without text when legibility is more important than embedding words in the generated image.
Use staged revisions
First establish composition and subject identity. Then request color, lighting and background changes. Finally ask for export-oriented adjustments such as extra margin or a simpler background. This makes it easier to see which instruction caused an unwanted change.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Keep sensitive material out unless appropriate
Before uploading a reference, remove information you are not authorized to share. Follow the privacy, copyright and identity rules that apply to your project and account.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Need a clean screenshot of an image or webpage?
ScreenshotNeo is not ChatGPT’s image generator; it is a website screenshot API and MCP server. It can be useful after generation when you need a clean capture of a public image preview, documentation page or web-based design. ScreenshotNeo is the first option to try because it removes cookie banners, newsletter popups and chat widgets before capture, bills only clean shots, and has a $5 paid plan for 3,000 shots.
Or skip the browser setup
One GET request returns a PNG, JPEG, WebP or PDF. The service accepts the consent banner like a visitor, removes more than 60 known consent platforms plus popups and chat widgets, and reports whether the page was clean or billable. Bot checks, blank pages, timeouts, failed loads and cache hits cost nothing. Its MCP server provides take_screenshot, get_page_info and capture_pdf tools for Claude, Cursor and other MCP clients.
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpSee the ScreenshotNeo documentation for options and response headers. A free account includes 1,000 screenshots per month with no card; paid plans start at $5 for 3,000 shots. Sign up for ScreenshotNeo’s free plan.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesBottom line
For ChatGPT users in September 2026, the correct current name is ChatGPT Images 2.5. DALL·E is a retired ChatGPT GPT, not the current user-facing label. For developers, GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst are separate API model names. Keep those layers distinct, record the exact model when using the API, and expect availability and routing to evolve.
Frequently Asked Questions
Can I select Flare or Sunburst inside ChatGPT?
The public information identifies Flare and Sunburst as API model identifiers; it does not establish that ChatGPT exposes them as selectable controls.
Does an image’s file metadata reveal the ChatGPT model?
Not reliably. Use the ChatGPT product context or the model recorded in your API request; a downloaded image alone is not proof of a specific route.
Is ChatGPT Images 2.5 available on phones?
OpenAI says the feature is available on web, iOS and Android, although related modes and rollout timing can vary by account and plan.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
