I tried ChatGPT Images 2.0: a fun, huge leap forward – and surprisingly useful for real work

Follow ZDNET: Add us as a favorite source On Google.

ZDNET Highlights

• Images 2.0 provides accurate text and usable graphics.
• It can match brand styles, including ZDNET visuals.
• Errors still arise, requiring human review.

Earlier this week, OpenAI unveiled its new image generation engine ChatGPT Images 2.0. Key to this release is a jump in functionality from creating “decorations” (OpenAI’s term) to full-page graphics including detailed text.

I had early access to the pre-release version. This worked quite well, but the ZDNET logo kept getting messed up. Now that the product has been officially released, I’ve been testing it extensively amid various challenges.

Imagination 2.0 is available for all ChatGPT tiers, but more capable language features are only available to paying tiers that can use Thinking Models. I’m running all these tests using a ChatGPT Plus account with Thinking turned on.

Also: I put GPT-5.5 through a 10-round test: it scored 93/100, losing points only due to excitement

Let’s start with the ZDNET branding exercise. Instead of just uploading the ZDNET page and finding the logo on the page, I created a standalone image of the ZDNET logo and uploaded that with each prompt. This helped a lot.

(A quick note: ZDNET does not allow OpenAI to scrape its pages. ZIDNET’s parent company, Ziff Davis, filed a lawsuit against OpenAI in April 2025, alleging that it infringed Ziff Davis copyrights in training and operating its AI system. So I used a Chrome extension to capture full-screen screenshots of the articles I wanted to test with Images 2.0. This way ChatGPT was able to read them.)

Can Images 2.0 protect the ZDNET logo?

My starting point was an article I previously wrote about Images 2.0. I gave ChatGPT this prompt: “Create a detailed and vivid infographic of this article using the ZDNET brand style and the attached ZDNET logo.”

Not only is the logo perfect, but the colors are perfect for ZDNET. But where the image really shines is the use of text. All text is correct, even small text at an angle to the image.

Can it create stylized sketchnotes?

Next, I decided to revisit the Sketchnotes challenge I gave to Google’s Nano Banana a few months ago. The task at the time was to create a sketchnotes version of the US Bill of Rights. Nano Banana did a great job with the images, but I had to try again and again to explain it in the right words. Read the article to see what obstacles I had to overcome.

Also: I Used Nano Banana 2 to Create Perfect Sketchnotes: 5 Lessons Learned

For ChatGPT Images 2.0, I raised the stakes a bit. I wanted Sketchnotes, but I wanted them in ZDNET’s branding style. I’m using branding style throughout this article because it’s the only way ChatGPT Images 2.0 can provide real value to users.

Here’s the first prompt: “Make a sketchnote of the US Bill of Rights for me. Use the ZDNET logo style and make the sketchnote in the ZDNET style.” That’s the image on the left. Here’s the second prompt: “Include the ZDNET logo and add more neon-style colors, perhaps on a black background.” That’s the image on the right.

First of all, make sure that the text is correct. There are no duplicates. Nothing is missing. Already, it is far above the performance of the Nano Banana. Both versions fit with ZDNET’s style. The only thing I’m not thrilled with is that the ZDNET logo looks jammed in the second image. Still, the logo is spot on, and I could probably do a few more quick passes to position it better.

Wacky Fun with an Infographic

But now we come to the unexpected error that my test set revealed. I asked Images 2.0 to turn my AI Website Builder Shootout article into an infographic. This produced a fairly useful, if somewhat busy, infographic. It even went to the internet and added information like base pricing that was missing from my article.

But there are four glaring errors:

1. The header highlights “Here are the 9 best AI website builders.” It also makes the “9” stand out. Except that only five website builders were reviewed. In the infographic below, it shows the five that I review. Oops.
2. The services I reviewed were Hostinger, GoDaddy, Wix, 10Web, and Squarespace. ChatGPT, for some reason, decided to replace 10Web with Durable (a competitor of 10Web). I did not review the durables. I didn’t even mention durable. Unique.
3. AI prepared a summary table for services, listing star ratings for ease of use, design flexibility, and AI features. But I did not provide star ratings for these categories. The AI ​​was overly generous towards some sellers, in a way that directly contradicted the review text. Odd.
4. Finally, and this is a knight, but still. At the bottom, there is a tilted line just above where the AI ​​has correctly reproduced the ZDNET logo. Why?

Also: Best AI Image Generator: Now There’s Only One Clear Winner

To be fair, these are all errors that a home human graphic designer can produce in a first draft. In my years as a founder and a product manager, I’ve definitely seen more serious graphics errors from my designers in their first drafts.

When I reprompted Images 2.0 with the corrections (except for the star rating, which I did not correct in the second image), it correctly modified the infographic with more appropriate information.

Chatgpt Images has come a long way

This ImageJ 2.0 release is a major improvement over previous versions. The ChatGPT Images version I saw last year was impressive, especially for re-contextualizing images.

Also: I got an early look at ChatGPT Images 2.0, and it’s impressive – with one exception

This new version, which can interpret actual content and then create images, is a huge leap forward compared to previous builds. More importantly, it can provide very solid professional value, making it very valuable not only for fun photos but also for real work.

Stay tuned as I’ll see how this build compares to the Google Gemini Nano Banana. I’ll pursue it even further to see what other work-related tasks it can help with, especially when it comes to user interface design.

How comfortable are you with trusting AI-generated visualizations, knowing that the model may introduce subtle factual errors? Let us know in the comments below.

You can follow my daily project updates on social media. Be sure to subscribe My weekly update newsletterAnd follow me on Twitter/X @davidgewirtzon facebook Facebook.com/DavidGewirtzon instagram Instagram.com/DavidGewirtzon bluesky @DavidGewirtz.comand on youtube YouTube.com/DavidGewirtzTV.