TL;DR. ChatGPT Images is OpenAI’s built-in image generator inside ChatGPT, running on the gpt-image-2 model. It replaced the aging, standalone DALL·E GPT as the default way to make and edit pictures in ChatGPT starting April 2026 — and the branded DALL·E GPT itself retires August 30, 2026 (OpenAI). Free tier included; no separate app.
Last reviewed: August 7, 2026. Reviewed quarterly — this is a fast-moving rename in progress.
If you searched “what is DALL-E” or “did ChatGPT stop making images” this week, you’re not alone. On July 31, 2026, OpenAI posted a one-line release note announcing that the official DALL·E GPT — the dedicated chatbot inside ChatGPT that made images through the old DALL·E 3 model — retires on August 30, 2026. The news traveled through AI-news accounts on X and lost its nuance in three or four retellings, until “the DALL·E GPT is retiring” quietly mutated into “ChatGPT stopped making images.” It didn’t. What actually happened is simpler and less alarming: OpenAI is retiring an old branded doorway because the main road, ChatGPT Images, got better and there’s no reason to keep maintaining two.
ChatGPT Images is the name for OpenAI’s native image generation and editing tool built directly into ChatGPT. You don’t search for it, install it, or open a special GPT — you just ask ChatGPT for a picture, in a normal conversation, and ChatGPT Images is the system that answers. It generates new images from a text description and edits images you upload, and since April 2026 it has run on OpenAI’s gpt-image-2 model, sometimes labeled “ChatGPT Images 2.0” in OpenAI’s own materials. In plain terms: DALL·E was a specific model wrapped in its own chatbot; ChatGPT Images is that same job, done natively, by the model that now powers the rest of ChatGPT.
What ChatGPT Images actually is, in plain language
ChatGPT Images is OpenAI’s answer to “make me a picture” or “edit this photo” typed straight into a ChatGPT conversation — no plugin, no separate app, no dedicated GPT required. It does two jobs: generating a brand-new image from a written description, and editing an image you upload by changing only the part you describe. OpenAI launched the current version, gpt-image-2, in its API and Codex on April 21, 2026, and it reached ChatGPT the next day as “ChatGPT Images 2.0” (OpenAI, April 2026). Unlike the old DALL·E GPT — a separate chatbot bolted onto ChatGPT specifically to run DALL·E 3 — gpt-image-2 is built directly into the same underlying model architecture that answers your text messages, which is why it can follow a multi-step conversation about one image instead of starting fresh every time.
The name itself is doing something interesting: it’s drifting from a specific product name into a generic description, the same pattern that turned “Google” into a verb. “ChatGPT Images” first meant one particular OpenAI release. Increasingly, in casual conversation and in search behavior, it just means “the thing that makes pictures inside ChatGPT” — whatever model happens to be running it that month. That generic drift is exactly why this term causes confusion when a specific branded piece (the DALL·E GPT) gets retired: people conflate the wrapper with the whole capability.
| Term | What it actually refers to |
|---|---|
| DALL·E | OpenAI’s original standalone image-generation model, launched 2021; DALL·E 3 (2023) is its last major version |
| DALL·E GPT | The dedicated chatbot inside ChatGPT built to run DALL·E 3 — retiring August 30, 2026 |
| ChatGPT Images | OpenAI’s current name for native image generation inside any ChatGPT conversation |
| ChatGPT Images 2.0 / gpt-image-2 | The specific model that powers ChatGPT Images since April 2026 |
Why ChatGPT Images matters right now
ChatGPT Images matters because it marks the moment OpenAI finished consolidating image generation into one actively developed system — and because a lot of casual users are only now discovering that “DALL·E” and “ChatGPT’s picture feature” stopped being the same thing months ago. A few numbers show the shape of the wave:
- Search demand is rising and still winnable. “chatgpt images” spiked to roughly 165,000 monthly U.S. searches in June 2026 before settling near 49,500 a month by August 2026 — still low keyword competition and climbing year-over-year (DataForSEO, August 2026).
- The old brand name hasn’t faded yet. “dall e” still pulls about 60,500 monthly U.S. searches, more than the current product name — but it’s a much harder term to rank for, a sign the audience searching it is often confused, not loyal (DataForSEO, August 2026).
- ChatGPT’s overall reach makes this a mainstream feature, not a niche one. ChatGPT crossed roughly 900 million weekly active users in early 2026, more than double the 400 million measured a year earlier (DemandSage, August 2026).
- Image generation is a meaningful slice of that usage, not an edge case — internal usage breakdowns put image creation at around 9% of what people ask ChatGPT to do, alongside coding, video, and audio generation (per OpenAI usage data reported by DemandSage, August 2026).
- The retirement deadline is real and dated. OpenAI’s own release notes confirm the DALL·E GPT goes away August 30, 2026, with guidance to download anything you want to keep before then (OpenAI Help Center, July 2026).
How ChatGPT Images actually works
ChatGPT Images works as a conversation about a picture, not a one-shot generator you feed a prompt into once and walk away from. You describe what you want — “a colorful birthday poster with a cartoon T-rex and the text ‘Emma’s 8th Birthday’” — and gpt-image-2 generates a first draft directly inside the chat. Because the model reasons through the request before rendering, it plans layout and text placement rather than guessing pixel by pixel, which is the single biggest reason text inside ChatGPT Images output is legible where DALL·E’s often wasn’t (OpenAI’s ChatGPT Images 2.0 announcement, April 2026; corroborated by AI/ML API, April 2026).
The more important shift is what happens next. With the old DALL·E GPT, asking for a change usually meant a fresh regeneration that only loosely resembled the last version — every “try again” was closer to a new roll of the dice. With ChatGPT Images, a follow-up message like “make the T-rex’s smile bigger and move the date to the bottom” edits the same image in place, keeping everything you didn’t mention untouched. That consistency extends to real photos: upload a picture of your living room, ask ChatGPT to “change the accent wall to deep forest green, keep everything else exactly as it is,” and you get back your actual photo with just that change — not a newly generated room that vaguely resembles yours.
Third-party specification roundups, cross-checked against OpenAI’s own announcement, put gpt-image-2 at native 2K output (4K available in beta), aspect ratios ranging from roughly 3:1 wide to 1:3 tall, up to 10 consistent variations generated in a single request, and meaningfully stronger multilingual text rendering than DALL·E 3 ever had (AI/ML API, April 2026; GetImg.ai, April 2026). One honest limit worth flagging up front: some users report a grainy, “noisy” texture on detailed or stylized requests, a known artifact of the current model that giving it more specific lighting and material detail tends to reduce but doesn’t fully eliminate.
Where ChatGPT Images shows up in real work
ChatGPT Images shows up anywhere someone needs a picture but doesn’t have a designer, photographer, or studio on call. The common thread across every row below is the same: a task that used to require paid software, a stock-photo budget, or a specialist now takes one written sentence and a few seconds inside a chat window people already have open. The table below maps the workplaces where it’s landed hardest; the profession sections further down cover the honest catch for each.
| Use case | Who uses it | What it replaces |
|---|---|---|
| Product photography for listings | Etsy and Shopify sellers | Photo studios, stock-image subscriptions |
| Ad and social campaign visuals | Marketers, content creators | Waiting on a design queue |
| Classroom worksheets and posters | Teachers | Template tools, clip-art bundles |
| Virtual staging for listings | Real-estate agents, flippers | $300–500/listing staging vendors |
| Event posters and invitations | Parents, small event planners | Generic template graphics |
| Editing a real, existing photo | Anyone with a phone camera | Basic Photoshop skills |
| Multilingual social graphics | International small businesses | Separate translation + redesign passes |
What this means for marketers and content creators
If you make marketing visuals, ChatGPT Images changes how fast a first draft happens. You no longer wait on a design queue for a social post mockup, an ad concept, or an infographic — you describe it, get a draft in seconds, and refine it with plain-language notes instead of rewriting your whole prompt each time. The legible in-image text is the headline feature for this job: you can finally generate a graphic with your actual headline spelled correctly, something DALL·E consistently mangled into gibberish letters.
The honest limit is brand consistency at scale. ChatGPT Images gets you 80% of the way fast, but pixel-perfect alignment with brand guidelines across dozens of assets, and knowing when an image reads “AI cheap,” is still a human judgment call. Treat it as a first-draft engine, not a design-department replacement. The AI Image Generation course covers prompt craft, style control, and the brand-consistency checks that separate a usable draft from generic AI output — two lessons are free.
What this means for teachers and educators
If you make classroom materials, ChatGPT Images is the design help you never had budget for. A worksheet cover, a vocabulary poster, a diagram for tomorrow’s lesson — describe it in plain sentences the way you’d explain it to a teaching assistant, and you get a usable draft in under a minute, no design software required. Because ChatGPT Images now renders legible text reliably, labeled diagrams and posters with real instructional content (not decorative gibberish) finally work.
The limit here is safety-driven, not technical: OpenAI applies content guardrails around generating images of real, identifiable people, including students, so some perfectly reasonable classroom requests involving real photos may get restricted. Stick to illustrated, non-photographic classroom visuals and you avoid the friction entirely. FindSkill’s Design Anything with ChatGPT + Canva course — free to start, all 8 lessons — walks through this exact workflow: describe it in ChatGPT, polish and print it in Canva.
What this means for small-business owners
If you run a small business, ChatGPT Images removes the single biggest cost barrier to decent product photos: a studio. Take a phone snapshot of your item, ask ChatGPT to put it on a clean white background or a styled countertop, and generate several variations in one request — since a batch of up to 10 variations counts as a single generation against your usage limit, that’s real efficiency for anyone managing dozens of listings.
Two limits worth planning around. First, marketplaces are tightening AI-disclosure rules — Etsy, Amazon, and eBay increasingly require flagging AI-generated or heavily edited product images, so build disclosure into your workflow rather than bolting it on later. Second, the photo still has to match the real item; an AI-prettied product shot that doesn’t match what arrives earns returns and bad reviews. The Product Photos with ChatGPT Images 2.0 course teaches the exact prompt-to-listing workflow, with two lessons free.
What this means for freelancers and solo sellers
If you’re a solo Etsy or Shopify operator, ChatGPT Images is the in-house photographer you couldn’t otherwise afford for every new SKU. The upload-describe-refine pattern works the same way for a handmade candle, a vintage find, or a print-on-demand mockup — and because edits apply to your actual uploaded photo rather than regenerating from scratch, you can dial in lighting and background across a whole product line without losing consistency between listings.
The catch is the same one every honest seller should lead with: it’s your reputation on the line if the photo oversells the product, and disclosure rules are only getting stricter, not looser. Used honestly — real item, AI-enhanced background and lighting, disclosed where required — it’s a genuine unlock for anyone running a shop solo. The AI for Etsy Sellers course covers the full shop workflow inside the ChatGPT app, two lessons free.
What this means for entrepreneurs
If you’re building a real-estate side business — flipping properties, managing a few rental listings, or offering staging as a service — ChatGPT Images turns a cost center into a near-zero-cost workflow. Upload a photo of an empty room, describe the furniture style that matches your target buyer, and get a staged version back in seconds instead of paying a $300–500-per-listing staging vendor or hiring a photographer for every property.
The limit is legal, not technical: staged and edited real-estate photos carry disclosure requirements in most markets, and a virtually staged photo must be clearly labeled so buyers aren’t misled — never use it to hide a real defect. Used as honest enhancement, it’s a genuine margin unlock for anyone running listings as a business. FindSkill’s Virtual Staging with ChatGPT course walks through the exact prompt workflow, MLS-compliant labeling included.
Common misconceptions about ChatGPT Images
ChatGPT Images spread through a confusing, half-explained news cycle rather than a clean official launch, so a handful of confident-sounding myths are traveling with it. Each one below changes how you should actually think about the tool, not just a technicality — get these wrong and you either miss a deadline scrambling to save images you don’t need to save, or you assume a paid upgrade is required when it isn’t.
“ChatGPT stopped making images.” No — this is the single most common piece of confusion driving searches for this term right now. What’s retiring on August 30, 2026, is one specific branded chatbot, the DALL·E GPT. Native image generation inside regular ChatGPT conversations, which is what most people already use, is unaffected and actively being improved, not shut down.
“ChatGPT Images is just DALL·E with a new name.” Not quite. DALL·E 3 hadn’t been meaningfully updated since 2023; gpt-image-2 is a different, actively developed model built directly into ChatGPT’s core architecture rather than bolted on as a separate tool. The rebrand reflects a real technical change, not just marketing.
“You need ChatGPT Plus to use it.” False. Image generation is available on the free tier, typically a handful of images per rolling 3-hour window, with no payment method required, and outputs are yours to use commercially either way. Plus mainly buys you higher volume plus two features — Thinking mode and in-generation web search — most casual users won’t miss.
“It replaced DALL·E overnight, out of nowhere.” No — ChatGPT Images has quietly been the default path since OpenAI shipped gpt-image-2 in April 2026. The August 30 retirement of the DALL·E GPT is the last step in a consolidation that’s been underway for months, not a sudden switch.
“All AI image tools are basically interchangeable now.” Not really. ChatGPT Images’ edge is that it’s conversational — you generate, describe a change, and refine inside the same chat you already use for everything else, without switching apps. Dedicated generators like Midjourney or Nano Banana still win on raw stylized image quality; ChatGPT Images wins on staying inside your existing workflow.
Related terms
ChatGPT Images sits inside a fast-moving cluster of AI-image and content-authenticity terms. If you’re trying to understand how it compares to rivals, or what happens to trust and disclosure once every platform can generate a convincing photo, these companion explainers are the natural next stop.
- Nano Banana — Google Gemini’s rival AI image model, built into the Gemini app
- Grok Imagine — xAI’s image and video generator inside Grok
- SynthID — Google’s invisible AI-image watermark, the rough equivalent of what OpenAI attaches via C2PA
- C2PA / Content Credentials — the cryptographic “receipt” standard OpenAI uses to mark ChatGPT Images output as AI-generated
- AI Slop — the low-effort AI content flood that faster, cheaper image generation can feed
See also
The links below go deeper into ChatGPT Images’ neighborhood — the courses that teach the profession workflows above, prompt skills you can copy directly, related glossary terms, and the hands-on blog tutorials this page draws from. Grouped by type so you can jump to whatever format fits how you learn.
Courses
- AI Image Generation — Generate images with AI: prompts, styles, brand-safe output
- Design Anything with ChatGPT + Canva — Describe it in ChatGPT, Canva designs it — free to start
- Product Photos with ChatGPT Images 2.0 — The prompt-to-listing photo workflow for Etsy and Shopify
- AI for Etsy Sellers — Run an Etsy shop end to end inside the ChatGPT app
- Virtual Staging with ChatGPT — Replace a $500/listing staging bill with an 8-lesson workflow
- AI Photo Editing — Edit, enhance, and restore photos with AI tools
- Canva AI Mastery — Canva’s Magic Studio AI tools, hands-on
- Canva AI for SMB Marketers — Canva AI for small-business marketing graphics
- Social Media Marketing with AI — Plan and produce social content with AI
- AI E-commerce Operations — AI across the e-commerce back office
- Photography Business with AI — Run a photography business with AI help
- AI-Powered UX Design — Design better user experiences faster with AI
- Interior Design with AI — AI interior design: layouts, palettes, and mood boards
- Google Gemini Mastery — Master Gemini end to end, image tools included
- ChatGPT vs Claude — Choose the right assistant for your work
- Prompt Engineering — Write prompts that reliably get what you want
Related terms in this glossary
- Nano Banana — Google Gemini’s free AI image and editing model
- Grok Imagine — xAI’s image and video generator inside Grok
- SynthID — The invisible watermark Google bakes into AI images
- C2PA — Content Credentials: the cryptographic receipt for a photo
- AI Slop — The low-effort AI content flood, and why it’s getting worse
- ChatGPT Connectors — How ChatGPT links to your other apps and data
- ChatGPT Work — OpenAI’s enterprise-focused ChatGPT tier
AI skills (prompt templates)
- Product Photography Generator — Studio-quality product shots from a phone photo
- AI Image Prompt Generator — Write prompts that work across any AI image model
- AI Photo Editing Master — Background removal, inpainting, upscaling, style transfer
- AI Art Prompt Crafter — Prompt craft for DALL-E-style and Stable Diffusion-style models
- Ghibli Style Portrait Generator — Studio Ghibli-inspired portrait prompt formulas
- AI Meme Generator — Meme formats and humor prompts for ChatGPT
- Consistent Character Designer — Keep one character consistent across images
- Children’s Book Illustrator — Consistent storybook scenes and characters
- Shareable Aesthetic Prompts — Scroll-stopping visual prompts for social platforms
- AI Action Figure Image Creator — The viral figurine-style image trend
Related blog posts
- ChatGPT Image Generator: Life After DALL·E (2026 Guide) — The step-by-step walkthrough this page’s research draws from
- Virtual Staging with ChatGPT Images 2.0: Realtor’s Guide — Can it replace a $300/listing staging vendor?
- ChatGPT Images 2.0 for Etsy & Shopify: Studio Photos Without a Studio — Solo-seller product photo workflow
- Canva AI 2.0 vs ChatGPT Images 2.0: Which Wins Your Marketing Stack? — The honest side-by-side for small-business marketers
- How to Tell If an Image Is AI-Generated — How to spot AI images, including ChatGPT Images output
- Best Free AI Image Generators — Ten generators ranked, including no-signup options
Profession hubs
- Learn AI for Small Business — AI playbook for small-business owners
- Learn AI for Freelancers — AI skills for freelancers and solo pros
- Learn AI for Entrepreneurs — AI for founders and entrepreneurs
- Learn AI for Teachers — AI playbook for educators
The bottom line
ChatGPT Images is not a new product — it’s the name for what ChatGPT’s picture-making has already become: native, conversational, built on an actively developed model instead of a three-year-old one wearing a special chatbot costume. The DALL·E GPT retiring on August 30, 2026, is a cleanup step, not a shutdown. If you’ve never used ChatGPT to make an image, or you’ve only ever used the old DALL·E GPT, the honest move is to stop looking for a special doorway and just ask ChatGPT directly — that’s been the main road for months, and it’s the only road after August 30.
Frequently asked questions
What happened to DALL·E in ChatGPT? The branded DALL·E GPT — a standalone chatbot dedicated to image generation — is retiring on August 30, 2026. Image generation itself is not going away: it continues inside every regular ChatGPT conversation, running on OpenAI’s newer gpt-image-2 model, known as ChatGPT Images. The name DALL·E is retiring; the ability to make images in ChatGPT isn’t.
Is ChatGPT Images the same as ChatGPT Images 2.0? Yes. ChatGPT Images 2.0 is the specific name OpenAI used when it relaunched the feature on gpt-image-2 in April 2026. In everyday use, people now just say ChatGPT Images to mean whatever image model currently powers ChatGPT.
Do I need ChatGPT Plus to use ChatGPT Images? No. The free tier includes image generation, typically a limited number of images per rolling 3-hour window, with no payment method required. ChatGPT Plus raises that ceiling and adds features like Thinking mode and in-generation web search for images.
Is ChatGPT Images better than DALL·E? On the specific complaints people had about DALL·E — garbled text inside images, weak photo editing, and inconsistency across regenerations — yes, ChatGPT Images is a real improvement, because gpt-image-2 is an actively developed model where DALL·E 3 was frozen since 2023. It isn’t universally better at everything; dedicated tools still win on raw stylized image quality.
Can ChatGPT Images edit a real photo, not just generate a new one? Yes, and this is one of the biggest upgrades over DALL·E. You upload a photo, describe the specific change, and ChatGPT Images edits that image directly — keeping everything you didn’t mention unchanged — instead of generating a new picture that only loosely resembles the original.
What happens to my old DALL·E GPT images after August 30? Images you made inside the official DALL·E GPT live in that GPT’s chat history. When it retires on August 30, 2026, OpenAI’s own guidance is to download anything you want to keep before the cutoff. Images made through regular ChatGPT conversations, not the DALL·E GPT, are unaffected.
Sources
- OpenAI — Introducing ChatGPT Images 2.0. https://openai.com/index/introducing-chatgpt-images-2-0/ (accessed 2026-08-07)
- OpenAI Help Center — ChatGPT release notes. https://help.openai.com/en/articles/6825453-chatgpt-release-notes (accessed 2026-08-07)
- OpenAI Help Center — ChatGPT Free tier FAQ. https://help.openai.com/en/articles/9275245-chatgpt-free-tier-faq (accessed 2026-08-07)
- Tom’s Guide — You have until August 30 to save your ChatGPT DALL-E images. https://www.tomsguide.com/ai/chatgpt/you-have-until-august-30-to-save-your-chatgpt-dall-e-images-heres-how-to-avoid-losing-them-forever (accessed 2026-08-07)
- AI/ML API — GPT Image 2: Release Date, Features, and Everything You Need to Know. https://aimlapi.com/blog/gpt-image-2-release-date-features-and-everything-you-need-to-know (accessed 2026-08-07)
- GetImg.ai — GPT Image 2: Official Release, Features & What’s New (2026). https://getimg.ai/blog/gpt-image-2-rumours-leaks-release-date-2026 (accessed 2026-08-07)
- DemandSage — ChatGPT Statistics (August 2026) — Latest Active Users Data. https://www.demandsage.com/chatgpt-statistics/ (accessed 2026-08-07)
- Wikipedia — GPT Image. https://en.wikipedia.org/wiki/GPT_Image (accessed 2026-08-07)
- Notebookcheck — DALL-E leaves ChatGPT on August 30: download your images first. https://www.notebookcheck.net/DALL-E-leaves-ChatGPT-on-August-30-download-your-images-first.1360522.0.html (accessed 2026-08-07)
- Windows Report — OpenAI will retire the official DALL·E GPT on August 30. https://windowsreport.com/openai-will-retire-the-official-dalle-gpt-on-august-30/ (accessed 2026-08-07)
- Crypto Briefing — OpenAI retires DALL-E image generation from ChatGPT. https://cryptobriefing.com/openai-removes-dall-e-chatgpt/ (accessed 2026-08-07)
- Apiyi.com — GPT-image-2 officially released: a complete beginner’s guide. https://help.apiyi.com/en/gpt-image-2-official-launch-beginner-complete-guide-en.html (accessed 2026-08-07)
