New AI Image Model
OpenAI releases ChatGPT Images 2.5 featuring a new sketch tool and faster generation.
Anonymous
Key Points
1
OpenAI said users generate more than 3 billion images each week across ChatGPT Images and the GPT-Image API models, highlighting the scale of demand for its image-generation tools.
2
The model is designed to render real-world information more accurately, support complex layouts such as transparent-background images, and preserve requested visual styles more consistently.
3
Users can invoke the drawing tool by typing “@Sketch” in a ChatGPT conversation, while comments can be placed directly on generated images to target specific edits.
4
OpenAI said GPT-Image-2.5 Flare will be the default API option for most applications, offering higher-quality images than GPT-Image-2 at 50% lower latency and supporting use cases including visual search, rapid prototyping, and high-volume generation.
5
The previous Images 2 release had introduced 2K output, multiple aspect ratios, web research for up-to-date information, non-Latin language support, and Instant and Thinking intelligence levels—features that provide context for the progression to Images 2.5.
OpenAI has released ChatGPT Images 2.5, a faster and sharper image-generation model that cuts creation time by up to 50% while delivering richer textures and more natural lighting. Silicon Angle reports the upgrade also improves precision during multi-step editing, letting users change specific elements without affecting the rest of an image. The rollout spans ChatGPT on web and mobile, plus ChatGPT Work and Codex.
The release includes a new Sketch feature that lets users draw directly in ChatGPT and use those drawings as visual references for image generation. Seeking Alpha highlights that OpenAI also introduced annotations and templates for posters and product images. The company revealed users generate over 3 billion images each week, underscoring massive demand for its tools.
Faster Generation and Sharper Visual Quality
Images 2.5 reduces generation latency by 50% compared to Images 2.0 while improving real-world accuracy and texture detail. Zamin notes the model better preserves key objects from uploaded reference images. It also handles complex layouts like transparent-background images and renders requested visual styles more consistently across multiple attempts.
KuCoin reports the upgrade sharpens details throughout generated images while maintaining faster speeds. The model now supports more precise multi-step editing workflows. Users can target specific changes without accidentally altering surrounding elements, even in compositions with many overlapping objects.
New Sketch Tool and Annotation Features
Users can now invoke the drawing tool by typing "@Sketch" in any ChatGPT conversation. Silicon Angle explains that sketches function as visual references for image generation. This lets creators quickly rough out ideas and have ChatGPT render them as polished final images.
Direct annotations on generated images pinpoint specific edits. KuCoin adds that templates for posters and product images streamline common use cases. Prompt sharing features let users collaborate and reuse successful creative workflows with teammates.
Two New API Models for Different Workflows
OpenAI added GPT-Image-2.5 Flare and Sunburst to its API. Unite.ai reports Flare serves as the default option for most applications, offering higher-quality images than GPT-Image-2 at 50% lower latency. Flare targets visual search, rapid prototyping, and high-volume generation.
Sunburst is designed for detailed creative workflows that require greater editing control and precision. The dual-model approach lets developers choose speed or depth based on their needs. Both roll out gradually to API users over coming weeks.
Sentiment
Mostly PositiveNEGATIVENEUTRALPOSITIVE
Publishers
30
Articles
29
Reach
59
A
AnonymousAuthor
