ChatGPT Images Review 2026: Pricing, Features and Verdict

What it is

ChatGPT Images is OpenAI's image creation and editing feature inside ChatGPT, plus the GPT Image models in the OpenAI API. In ChatGPT, you ask for an image in any chat or open Images from the sidebar. You can upload a photo and describe changes, select part of an image to edit, start from a template, or, on mobile, sketch an idea and have it turned into an image.

The current version is ChatGPT Images 2.5, launched in ChatGPT on September 8, 2026. OpenAI says it brings sharper details, more precise editing and faster generation. In the API the matching models are gpt-image-2.5-sunburst, which OpenAI recommends when editing precision matters most, and gpt-image-2.5-flare, for fast high quality everyday generation. DALL-E has been retired from ChatGPT.

Pricing explained

There is no separate image subscription. Image generation comes with ChatGPT.

Plan	Price	Image generation
Free	$0	Limited image creation
Go	$8 per month (may include ads)	More image creation
Plus	$20 per month	More complex and accurate image creation, plus images with thinking
Pro	$100, $200 or $500 per month	Unlimited and faster image creation, subject to abuse guardrails
Business	$25 per seat monthly, $20 annually	Images with thinking; business data not used for training by default
Enterprise	Contact sales	Images with thinking coming soon, per OpenAI

Plan wording from ChatGPT's pricing page and help center in October 2026.

What "limited" means. OpenAI does not publish image counts per plan. Limits can vary with demand, and image generation shares your plan's usage. If you need a predictable number for a budget, use the API, which bills per image.

Images with thinking. On Plus, Pro and Business, ChatGPT can reason before it draws, which helps with complex layouts, infographics and multi part instructions. Free and Go do not get it.

API pricing. OpenAI's image guide lists per image prices for GPT Image 2 at 1024 x 1024: about $0.006 at low quality, $0.053 at medium and $0.211 at high. Portrait and landscape images at 1024 x 1536 cost about $0.005, $0.041 and $0.165. GPT Image 2.5 bills by token: $8 per million image input tokens, $2 cached, and $30 per million image output tokens, with Batch at half price. Partial images when streaming cost an extra 100 output tokens each.

Worked example. A store that needs 500 product scenes a month at medium quality would pay about $26.50 on GPT Image 2 by API, before text input tokens. The same volume on ChatGPT Plus has no published guarantee, so the API is the safer bet for steady production.

Compare with every rival in the Price Index and see the full plan list in ChatGPT pricing.

Key features

Conversational editing

The core advantage: you refine an image the way you would brief a designer. "Make the background warmer," "move the logo to the top right," "change the text to say Spring Sale." ChatGPT keeps the context of the conversation, so iteration feels like one continuous edit rather than new prompts.

Selection tool editor

Open an image, use Select to highlight an area, and describe the change. OpenAI's help center warns that highlights are not always precise and edits can extend beyond the selection. On mobile you also get Resize and Remove.

Text, transparency and aspect ratio

ChatGPT Images can add text, add details within the image and make the background transparent. You can generate in any aspect ratio with the picker or by asking. In the API, transparent backgrounds need PNG or WebP output.

Templates and sketch to image

New with Images 2.5: templates for posters, logos and other formats that ask follow up questions, and on mobile, a Sketch tool you draw on, then describe. You can also share a prompt template so someone else can make their own version.

Image library

Every image you create is saved under Images on web and mobile. To delete one, you delete the conversation it came from.

API controls

The API supports quality settings from low to max on GPT Image 2.5, custom sizes up to 3840 px per edge, masks for inpainting, multi turn editing through the Responses API, streaming partial images, and a moderation setting with standard or low strictness.

Limitations OpenAI lists

OpenAI's own documentation names four: complex prompts can take up to 2 minutes; text placement and clarity can still struggle; consistency for recurring characters or brand elements can slip across generations; and precise placement in layout sensitive compositions can be difficult. If you need the same mascot across 40 ads, test that before committing.

What users say

ChatGPT has the largest review base of any product in this category: 4.8 on the App Store from over 11.4 million ratings and 4.6 on G2 from 3,109 reviews, though those cover ChatGPT as a whole, not images alone. Trustpilot shows 1.3 from 1,260 reviews for openai.com, driven by billing, account and support complaints. Common praise for images: it understands plain language, edits are fast, and text is usually legible. Common complaints: hitting unexplained limits, content filters blocking harmless requests, and drift when asking for many small edits.

Commercial rights and copyright

From OpenAI's Terms of Use:

• Ownership. "As between you and OpenAI, and to the extent permitted by applicable law, you (a) retain your ownership rights in Input and (b) own the Output. We hereby assign to you all our right, title, and interest, if any, in and to Output."
• Not unique. Output may not be unique, and the assignment does not extend to other users' similar output.
• Restrictions. You may not represent output as human generated when it was not, or use output to develop models that compete with OpenAI.
• Responsibility. You must have rights to anything you upload, and you are responsible for content not violating the law or the terms.

OpenAI publishes no consumer IP indemnity for ChatGPT image output. Whether an AI image can be registered for copyright is a separate legal question that depends on your country and how much human authorship is involved.

Privacy and security

OpenAI's terms say you can opt out of having your content used to train its models. Business and Enterprise do not train on business data by default, and Business adds SAML SSO and MFA. Every image from ChatGPT, Codex or the API carries C2PA Content Credentials and an invisible SynthID watermark, per OpenAI's provenance page, and anyone can check a file at openai.com/verify. For linked teen accounts, parents can control whether the teen can create or edit images.

Who it is for

• Most people and small businesses who want good images with no learning curve.
• Marketers making social graphics, ads and thumbnails with text.
• Existing ChatGPT subscribers, since images are already included.
• Developers who want a well documented image API with editing and masks.

Who should skip it

• Artists who want a signature style and taste controls. Midjourney does that better.
• Heavy users on a budget who need predictable volume. Leonardo.Ai and Midjourney Standard offer unlimited slow queues.
• Brands that need indemnified, commercially safe training data. Look at Adobe Firefly.
• People who want the best free option. Nano Banana in the free Gemini app is the closer call; see ChatGPT Images vs Nano Banana.

Alternatives

• Nano Banana: Google's free Gemini image model with cheaper API rates.
• Midjourney: stronger art direction.
• Ideogram: text heavy layouts, print and logos.
• Adobe Firefly: includes GPT Image as a partner model alongside Adobe's own.

Verdict

ChatGPT Images is our top pick for most readers: it is on every plan including Free, conversational editing makes it the easiest tool to get a usable image from, Images 2.5 added templates and sketch input, and the API is documented down to the cent per image. It loses points for unpublished limits and for the consistency and layout weaknesses OpenAI itself lists. If you already pay $20 for ChatGPT Plus, start here before buying a separate image subscription.