The specifics:
The model can now maintain faces, lighting, and composition throughout revisions, and Image 1.5 offers noticeably faster generating times.
When compared to GPT-image-1's numerous problems, text rendering also much improves, with 1.5 managing lengthy content, infographics, and a variety of letter sizes.
The new model takes the top spot on the text-to-image and editing leaderboards for both Artificial Analysis and LM Arena.
Additionally, OpenAI introduced a new dedicated creative panel that joins the standard chat-based workflow to provide users with curated style options and quick-start templates.
When compared to GPT-image-1's numerous problems, text rendering also much improves, with 1.5 managing lengthy content, infographics, and a variety of letter sizes.
The new model takes the top spot on the text-to-image and editing leaderboards for both Artificial Analysis and LM Arena.
Additionally, OpenAI introduced a new dedicated creative panel that joins the standard chat-based workflow to provide users with curated style options and quick-start templates.
Even though GPT-image-1 was a huge success at the time, it doesn't take long to lag behind the cutting edge, thus this new update is long overdue. Like the most recent GPT-5.2 release, 1.5 meets the benchmarks and puts OpenAI at least on pace with Google's highly anticipated releases for the upcoming year.