ChatGPT's new image tool is up to 50% faster and edits better

OpenAI has released ChatGPT Images 2.5, a new version of the model behind ChatGPT's image tool. The company says it is up to 50% faster than before and makes more accurate edits. OpenAI says people create over 3 billion images every week with its image tools.

What changed

Compared with Images 2.0, OpenAI says the new model produces more natural lighting and richer textures. It keeps a person or object recognizable when working from reference photos. It also follows complex instructions, layouts and styles more closely.

Editing is the biggest gain, according to OpenAI. Earlier changes stay in place across several rounds of tweaks, and quality does not slowly fade the way it can with repeated edits.

Two versions for businesses

• Flare is the default. OpenAI says it beats GPT-Image-2 on quality at half the wait time. It suits social posts, prototypes and high-volume work.
• Sunburst is for polished work such as campaign images. It gives tighter control over edits but takes longer.

Availability and open questions

The model is rolling out to ChatGPT, ChatGPT Work and Codex users on all tiers, across desktop, mobile and web. It is also available to developers through OpenAI's API, the connection that lets other apps use its models.

OpenAI says it checks prompts and images to prevent harmful results. It also adds hidden watermarks and labels so images can be identified as made with its tools. Pricing and exact rollout dates were not given. The source mentions four new features inside ChatGPT but does not list them.