Draft poster · the finished cut will appear here
What happened
OpenAI says Images 2.5 is better at preserving reference subjects, lighting, and textures, and follows edits more reliably across multiple turns. OpenAI also reports up to 50 percent lower image-generation latency than Images 2.0. The release reaches ChatGPT and the API, with different model names for different creative workflows.
How it works
The product changes are as important as the model claim. OpenAI’s Sketch feature lets a user draw directly in ChatGPT and use the drawing as a visual guide. Templates provide a starting structure for formats such as posters or product images. Comments can point at a specific part of an image, and a shared image can optionally carry its prompt so another person can adapt the idea.
For developers, OpenAI describes GPT-Image-2.5 Flare as the faster default choice for most applications. Sunburst is positioned as the slower, more controlled option for high-precision creative work. The API therefore exposes a tradeoff between speed and control instead of treating every generation as the same kind of job.
Why it matters
The useful upgrade is the interface around the model. A sketch or a comment can express spatial intent directly, while multi-turn consistency protects decisions that have already been made. That is closer to manipulating a visual tool than explaining every revision in prose.
What we don’t know yet
OpenAI’s latency and quality comparisons are product claims, not independent benchmarks. Availability, defaults, pricing, and the behavior of the examples can change; the release does not establish a universal quality advantage.
Source: OpenAI, OpenAI, September 8, 2026 - original post