- Learn Prompting's Newsletter
- Posts
- The New "ChatGPT Images" is a Game-Changer
The New "ChatGPT Images" is a Game-Changer
With precise editing, reliable text, and 4x faster speeds, it's finally moving from a fun toy to a practical creative tool.
Learn Prompting Newsletter
Your Weekly Guide to Generative AI Development
OpenAI's New Flagship Image Model: GPT-Image 1.5
Learn about its precise editing, reliable text, and 4x faster speeds
Hey there,
We’ve all been there. You generate an almost perfect image, but the text is garbled or a small detail is wrong. You try fixing it with a new prompt but end up with an entirely new (and completely wrong) image. This frustrating experience is exactly what OpenAI is addressing with their new flagship image model, GPT-Image 1.5.
The Main Upgrades
GPT-Image 1.5 isn’t bringing just a few minor improvements, it's a major upgrade focused on three key areas.
Precise Editing:
ChatGPT is now able to make small, focused edits that only impact the elements you specifically point out in your directions. This gives you the power to correct any small issues without needing to completely restart your creation process. So now you won’t have to worry about the rest of an image changing when you’re just trying to fix one small detail. The iterative process is also made significantly better by images generating 4x faster.
You can now seamlessly change your images in seconds.
More Reliable Creation and Text Generation:
OpenAI’s new image model “follows instructions more reliably”. This upgrade means both your edits and initial creations will better reflect your prompts. GPT Image 1.5 is also capable of generating clear, legible text within images. This has been something that many image models have struggled with in the past.
Here I uploaded a past newsletter and asked ChatGPT to add it to an image.
A New Creative Space:
In addition to the new model, ChatGPT has also received a new dedicated “Images” section in ChatGPT. This space is designed to help ease the creation process by offering example images and prompts. You can also see and edit all your ChatGPT created images on this page.
The “Images” section offers new styles and example prompts to experiment with.
Access & Availability
In ChatGPT: The new model and Images experience are live for all users (including free). You can access both via the ChatGPT app and website.
In the API: The new model is available as GPT-Image-1.5. This new model is 20% cheaper than GPT-Image-1.
We’re now hosting enterprise AI workshops for teams, starting at $12K, focused on responsible and effective AI use in business.
Contact Us →We’re also opening enrollment for our AI Red-Teaming Masterclass — designed for professionals in cybersecurity, DevOps, and app security, as well as technical leaders like CTOs and CISOs looking for hands-on AI security experience. The course is $1,199 per student with discounts available for bulk seats.
Enroll Now →My Thoughts
I’ve always thought image generation is one of the most interesting and impactful use cases for AI. Giving people the ability to create unique visuals for work, school, or personal projects is incredibly powerful. However, previous AI image tools felt very limited in their scope and editing abilities. Luckily, models like GPT Image 1.5 help turn image generation models into more than just cool experiments. For me, the new editing and text generation abilities are the standout improvements here. I can still remember having to create dozens of images just to get one good output that correctly included three or four words. The new “Images” section is also a welcome addition to ChatGPT and another clear step in OpenAI’s strategy to make ChatGPT into a one-stop-shop for every AI use case.
The new Images experience has rolled out to everyone! I highly recommend jumping in and trying the new editing features.
Reply