- كيفية الاستخدام: Just type what you want in a normal sentence, like “draw a red fox reading a book by a window, watercolor style.” ChatGPT makes the picture inside the chat.
- To edit: Reply with what to change (“make the sky sunset orange”) or upload your own photo and describe the edit.
- Plans: Free works but is limited and slower. Go and Plus allow more images and more detail. Pro is unlimited within OpenAI’s abuse guardrails and faster.
- Free alternatives exist if you hit limits — see our free AI image generators guide.
The ChatGPT image generator is built into ChatGPT itself. You don’t need a separate app or a special mode. You open a normal chat, describe the picture you want in plain words, and ChatGPT draws it. If you want changes, you reply in the same chat and say what to change. That’s the whole idea.
This guide walks a first-time user through it: how to ask, how to edit, what each plan allows, and how to write prompts that actually give you the picture in your head.
How to Make an Image in ChatGPT
Open ChatGPT on the web (chatgpt.com) or the mobile app. Start a new chat and type your request as if you were asking a person. For example:
“Make me a picture of a golden retriever puppy sitting on a picnic blanket in a park, soft afternoon light, in the style of a children’s book illustration.”
ChatGPT will generate the image inside the chat, usually in under a minute. Tap or click the picture to see it larger, then use the download button to save it.
A quick note on wording. In AI, a “prompt” just means the instruction you type. A “model” is the AI system doing the work. You don’t have to pick a model for images — ChatGPT chooses one for you.
How to Edit an Image You Already Have
You can upload a photo and ask ChatGPT to change it. Click the paperclip or plus icon in the message box, choose your photo, and then type what you want done. For example:
- “Remove the person in the background.”
- “Change the wall color from white to sage green.”
- “Turn this into a pencil sketch.”
- “Make this photo look like it was taken at golden hour.”
ChatGPT returns a new image. If it’s not quite right, just reply with more instructions in the same chat: “Keep everything else the same but make the sky cloudier.” You don’t have to start over.
If your goal is to fix a blurry or low-quality photo rather than restyle it, a dedicated tool often works better — our guide on how to unblur a picture covers free options.
What Each ChatGPT Plan Allows
The image generator works on every plan, but how many images you can make and how detailed they can be depends on which plan you’re on. These figures were verified on chatgpt.com/pricing on 11 September 2026.
| الخطة | توليد الصور | Fast-model context | Reasoning context |
|---|---|---|---|
| مجاني | Limited and slower | 27K tokens | — |
| Go | More images than Free | 54K tokens | ٢٥٦ ألف رمز |
| بالإضافة إلى ذلك | More complex and detailed images | 54K tokens | ٢٥٦ ألف رمز |
| برو | Unlimited and faster, within abuse guardrails | 128 ألف رمز | 400K tokens |
A quick term: “context” is how much text ChatGPT can hold in mind at once during your conversation. A “token” is roughly three-quarters of a word. Bigger context helps with long chats and complex edits where you keep refining the same image.
OpenAI does not publish exact daily image counts on the pricing page — the limits shift and are described as “limited,” “more” and “unlimited within abuse guardrails.” If you’re bumping into caps on Free, the practical fix is Go or Plus. For a full plan comparison, see our breakdown of ChatGPT Free vs Go vs Plus vs Pro. The official page is openai.com/chatgpt/pricing.
One important warning: the numbers you sometimes see in the news about huge context windows — a million tokens or more — come from the API, which is the developer version. Those figures do not apply inside the ChatGPT app. What you get in ChatGPT is the table above.
How to Write a Prompt That Actually Works
Vague prompts get vague pictures. The trick is to cover five things: subject, style, setting, lighting, and shape.
- Subject: What is in the picture? Be specific. “A tabby cat with green eyes” beats “a cat.”
- Style: Photograph? Oil painting? Cartoon? Pixel art? Watercolour? Say it.
- Setting: Where is the subject? “On a wooden desk beside a coffee cup” gives ChatGPT something to build.
- Lighting: “Soft morning light,” “harsh midday sun,” “neon glow at night.” Lighting changes the whole mood.
- Shape: Ask for square, portrait (tall) or landscape (wide). If you need it for Instagram, say square. For a phone wallpaper, say portrait.
A weak prompt: “a house.” A strong prompt: “A small stone cottage with a red door and ivy on the walls, set in a misty English countryside at dawn, soft golden light, painted in a storybook illustration style, landscape shape.”
If the first result is close but not right, don’t rewrite from scratch. Reply with a targeted change: “Same picture but move the door to the centre.” For a longer walkthrough with examples, our beginner’s guide to AI images is a good next step.
What ChatGPT Will and Won’t Draw
ChatGPT refuses some requests. In general it won’t generate images of real named people in situations they didn’t consent to, sexual content, graphic violence, or things that mimic copyrighted characters closely. Rules change over time and vary by country, so rather than list them here, check OpenAI’s سياسات الاستخدام for the current version.
On usage rights: OpenAI generally lets you use images you create for personal and commercial purposes, but the details (and any restrictions on likenesses, logos or trademarks) are laid out in their terms. Read them if you plan to sell or publish what you make.
A short honesty note. AI image tools sometimes get hands, text on signs, and small details wrong. They can also produce something that looks convincing but is factually off — a “map of Italy” with the wrong coastline, for instance. Treat the output as a draft, not a source of truth. And please don’t use these tools to impersonate real people or pass off AI images as photographs where that would mislead someone.
When ChatGPT Isn’t the Right Tool
ChatGPT is convenient because the image generator sits inside the same chat where you write, plan and brainstorm. But it’s not always the best fit.
- You’ve hit the Free limit. Rather than waiting or upgrading, try a free alternative. Our roundup of أدوات توليد الصور بالذكاء الاصطناعي المجانية lists real tools with real free tiers.
- You need many variations fast. Dedicated image tools often let you generate four or more options per prompt in a single click, which is faster than asking ChatGPT one at a time.
- You’re editing photos of yourself. For portraits and profile pictures, a specialised service usually does a better job. See كيف تحصل على صور رأسية بالذكاء الاصطناعي.
- You want to reverse-engineer a picture you like. The Convly image-to-prompt tool reads an image and suggests a prompt you can paste back into ChatGPT.
Windows, macOS, Linux and Mobile
ChatGPT’s image generator works the same on every device because it runs in the cloud. On ويندوز و macOS you can use the ChatGPT website in any browser, or install the official desktop app from openai.com. On لينكس there’s no official desktop app, but the website works in Firefox, Chrome or any modern browser. On iPhone و Android, use the official ChatGPT app from the App Store or Google Play — image generation and photo upload both work from mobile, including from the camera roll.
The one thing to watch: on mobile, uploaded photos sometimes get compressed, which can reduce edit quality. If detail matters, upload from a computer.
الأسئلة الشائعة
Can I make images on the free ChatGPT plan?
Yes. The free plan includes image generation but with limits on how many you can make in a day and slower speed at busy times. If you run out, you either wait until the limit resets or upgrade to Go, Plus or Pro. The exact daily count isn’t published — OpenAI describes Free as “limited” on its pricing page.
Can I use ChatGPT images for my business?
Generally yes, based on OpenAI’s terms, but there are important exceptions — for instance, images of real people, trademarks and some copyrighted styles. Before publishing or selling anything, read OpenAI’s سياسات الاستخدام and terms of use. Rules change, and a lawyer’s view beats a guide’s summary if money is on the line.
Why does ChatGPT sometimes refuse my image request?
It refuses when a request bumps into its safety rules — real people, explicit content, certain trademarks, or things that look like real IDs and currency. Sometimes the refusal is over-cautious. Rephrasing helps: describe the scene generically rather than naming a specific person, brand or character.
Is ChatGPT’s image generator better than Gemini or Claude?
They’re close, and the “best” one changes with each update. ChatGPT tends to be strong at illustration and editing with follow-up instructions. Gemini is tightly linked to Google’s other tools — see Google AI plans explained. For a head-to-head on the text side, our Claude مقابل ChatGPT comparison covers the differences.
Can I use ChatGPT to remove backgrounds or fix blurry photos?
You can ask it to, and it will try, but dedicated photo tools usually give cleaner results for those specific jobs. For blur and quality, see our unblur guide. For general photo editing, the best AI photo editing apps lists purpose-built options.
Does ChatGPT save my images or share them?
Images you generate stay in your chat history and in your account. OpenAI may use conversations to improve its systems unless you turn that off in settings (Data Controls). If a picture is private or sensitive, either turn off training data use before generating, or use a plan and setting that excludes your content from training. Check the current details on OpenAI’s privacy page.