Get Your Money's Worth from ChatGPT Images 2.0 (2026)

ChatGPT just released a new image generation model, GPT Image Gen 2, and it generates a lot more than pretty pictures. In this walkthrough I take it through four real-world use cases: product mockups, infographic generation with thinking mode, before and after visualization for landscaping and repairs, and progress timelines. If you run a business, especially a brick-and-mortar one, this is for you.

What you need

• A ChatGPT account with access to image generation
• The Create an image option in a new chat; that is the whole setup
• A product, a property, or a service you want to visualize

1. Where image generation stands right now (0:12)

Image generation was dominated by Google with Nano Banana Pro and Nano Banana 2 for roughly the last five to six months. Three days before I recorded this, ChatGPT released GPT Image Gen 2, and with both players pushing, 2026 is going to get very interesting for anyone who sells something physical.

2. How to access it, and the thinking tag (0:44)

Click Create an image in ChatGPT. You will see several explore idea options, and some carry a thinking tag: those take longer because the model fetches information from the web to give you the most accurate result. You can also change the resolution. By default it selects the best resolution for the use case, but if you want a specific format for LinkedIn, Instagram, or TikTok, you can simply force it. Standard generations take around thirty seconds to one minute; thinking makes it longer.

3. Use case 1: product mockups (1:20)

I invented a supplements brand on the spot and asked for a mockup:

Hey, I am just creating a new brand of supplements called Galactic Supplements. What I need you to do is I need you to generate a product mock-up with white background so that I can use it.

The result: Galactic Supplements, nice and shiny, and no problems with the text on the label, which used to be the giveaway flaw in AI product shots.

4. Use case 2: infographics with thinking mode (2:48)

For this one I used thinking mode, where you can select either standard or extended: the model reasons a lot before producing the final image. The prompt is one of the stock suggestions, so you can reproduce it exactly:

Create a visually rich infographic about an endangered animal.

I got a leopard infographic with the habitat, the diet, why it is at risk, unique traits, and fast facts. Zoom in and the quality holds: no broken icons, no broken text. You could generate an entire book of animals or plants this way and assemble the images into a PDF, or use the same trick for portfolios.

5. Use case 3: before and after for properties (4:15)

I generated a satellite view of a house with this prompt:

I want you to create a satellite view of a house with a big patio behind it.

Then, in the same Create an image flow, the edit:

Hey, now can you now add a pool to that patio, please? Make sure that it's quite close to the white sofa in the left-hand corner.

And there you go: pool in the backyard, exactly where I asked. Any company that needs to show the end result before work starts, in landscaping, construction, or property, can now help clients project themselves onto what they are investing in.

6. Use case 3b: repair quotes and DIY guides (5:12)

Same logic for repairs: I took an AI-generated cracked wall and asked for before and after pictures of the fix at three different price points. If your business does infrastructure repairs, you can even add your own pricing table to the image. And if the client wants to fix it themselves, ask for a step-by-step guide: mine came back with clean the area, apply spackle, sand smooth, prime, and paint, with the tools needed and the given budget.

7. Use case 4: progress timelines (6:12)

For people like me who do not have a lot of hair: the medical world will likely use this to show how progress looks if you stick to a plan. Think workout routines, hairdressers, hair transplant specialists, a timeline generated with the actual face of the person. One honest caveat: I do not know how such an image is legally binding for a business, so treat this one as delicate for now.

8. Bonus: UI screenshots (6:55)

The model is also unbelievably powerful at generating user interface screenshots, with the correct logos, correct spacing, and correct bars. Useful for mockups and presentations without touching a design tool.

Pitfalls and tips

• Thinking mode takes noticeably longer. Use it for infographics and information-dense images, skip it for simple shots
• Force the resolution when the image is destined for a specific platform instead of relying on the default
• Be careful with before and after images in client-facing quotes; the legal weight of an AI projection is still an open question
• Iterate with follow-up prompts in the same flow, like the pool edit, instead of regenerating from scratch

Where to go next

To run image and video generation at campaign scale from a single Claude conversation, watch Claude + Higgsfield: AI Marketing Assets in Minutes. And to keep every generated asset consistent with your brand, read Design Systems for AI: The 2026 Guide.