ChatGPT’s image generation model just received a significant upgrade, going from DALL-E 3 to GPT Image 1.5 Hazel.
OpenAI began the phased rollout for the GPT 5.2-powered image model on December 11. The official release was available as of December 16.
This upgrade came less than a month after Google upgraded Gemini’s image model to Nano Banana Pro.
Introducing ChatGPT Images, powered by our flagship new image generation model.
— OpenAI (@OpenAI) December 16, 2025
– Stronger instruction following
– Precise editing
– Detail preservation
– 4x faster than before
Rolling out today in ChatGPT for all users, and in the API as GPT Image 1.5. pic.twitter.com/NLNIPEYJnr
New Image Generation Features with ChatGPT
GPT Image 1.5 has exciting features and several improvements compared to DALL-E 3, including:
Improved Speed
Users can expect image generation to be four times faster than the last model. Not only are image generations quicker, going from roughly a minute or longer to mere seconds, but users will also notice less lag during peak periods.
New Types of Editing
ChatGPT is now better capable of changing the color of objects, accurately adjusting the lighting, repositioning objects, and erasing people or elements from a scene. It can also effectively add effects, like blur or shadows, to a photo.
Enhanced Text Rendering
Text rendering has also been improved with this update. Now, images generated with text have enhanced accuracy, fewer spelling errors, and are more legible—making generations cleaner and more reliable than ever before.
The model is more proficient when it comes to adding dense and small text to both existing and AI-generated images. This can especially come in handy when creating infographics, event invitations, product mockups, educational materials, menus, and more.
More Reliable Instruction Following
No more getting frustrated when using ChatGPT to edit pictures or generate images from scratch. Say hello to improved instruction following, which means fewer prompt rewrites, less time wasted, and more reliable generations upon the initial ask.
GPT Image 1.5 is better at handling multiple instructions, keeping editing within context, and creating images using its enhanced reasoning capabilities. It’s also less likely to use common stereotypes when creating outputs.
Adult Content Policy Update
Users aged 18 and older will be able to opt in to generate adult content (via both text and images) after they undergo a robust age verification process.
Adult content includes erotic conversations, sexually explicit image generations, and the creation of mature creative pieces, like fan fiction.
The latter are still banned for unverified users and those under age 18. ChatGPT also still maintains strict prohibitions on non-consensual content, content involving minors, hate speech, and harassment, regardless of age.
Other New Additions
- Higher resolution output
- Batch generation capabilities
- Improved character consistency
- Enhanced photo realism
- Billions of style presets
- Virtual clothing and hairstyle try-ons
- Different generation settings depending on whether you prefer speed or high-quality results
These upgrades are impressive considering past ChatGPT image generation models faced several limitations and weaknesses—including inaccurate text rendering, slow image generation speed, and limited customization options.
What People Have to Say
There has been both excitement and backlash regarding GPT Image 1.5.
“…OpenAI released GPT Image 1.5 Hazel, and early results suggest it is closer to beating Nano Banana Pro than anyone expected…” said PROFIT. “This isn’t about pretty images anymore.
It’s about high-utility, multimodal visual intelligence.”
Others, like Brandtech Group’s Generative AI Creative Director, James Dow, are not likely to find the update valuable.
“It will be another fantastic tool in the box, but it’s not a killer of anything… It’s not going to reinvent any existing workflows for me yet,” he said about GPT 4.o (Omni), which was released on May 13, 2024.
GPT 1.5 vs Gemini’s Nano Banana Pro
Both ChatGPT and Gemini’s recent image generation upgrades have achieved impressive improvements. However, each has its own strengths and weaknesses.
GPT 1.5 primarily excels with instruction following, deep reasoning, and a versatile creative range. Meanwhile, Nano Banana Pro’s strong suits are in raw image quality, multi-image consistency, and real-time data integration.
While Nano Banana Pro is still superior when it comes to generation speed, text rendering, higher resolution, and character consistency, the freshly released GPT 1.5 exhibits greater non-literal concept integration, prompt adherence, logical spatial relationships, and API flexibility.
Who Has Access to GPT Image 1.5?
The new update to GPT’s image model is available to all users, including free users.
However, unlike free users, Plus, Team, and Pro subscribers have higher limits for image generation. When free users reach their daily or hourly limit, they will temporarily revert to an older image generation model.Interested in seeing what GPT Image 1.5 has to offer? Access the website or app to get started with image editing or generation today.
Optimize Your Brand’s Visibility in AI Search
Millions turn to AI platforms daily to discover and decide. Make sure they find your brand.
