OpenAI’s ChatGPT Images 2.5 is a newly released upgrade to its image generation platform designed to support faster, more precise visual content creation and editing. Unveiled on September 8, 2026, the tool introduces improvements such as sharper and more natural lighting, richer textures, consistency across edits, and new creative controls like the Sketch feature, templates for popular formats, and the ability to leave comments directly on images. It is available to all ChatGPT, ChatGPT Work, and Codex users on desktop, mobile, and web, and comes with new models in the API—GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst—for enhanced performance and precision.
Key Features
Image fidelity and recognition: Images 2.5 improves on earlier versions by better preserving subjects (such as people or pets) when transforming reference photos, rendering lighting and textures more realistically, and following editing instructions more reliably.
Editing precision: Users can now make edits that affect only specified parts of an image (background, copy, objects), leave comments on images to guide refinement, and maintain visual consistency across multiple revisions.
Creative input tools: A new Sketch tool lets users draw directly in ChatGPT as input for final images; templates cover formats like posters and product photos to streamline layout work.
Speed and scalability: Image generation latency has been reduced by up to 50% compared with Images 2.0, helping ideation and refinement move more quickly. Two API models—Flare and Sunburst—offer options balancing speed and maximum detail.
Pricing
Business‐oriented pricing for ChatGPT includes two seat types: Standard and Premium. Standard seats cost $20 per user per month when billed annually ($25 if billed monthly), and Premium seats cost $100 per user per month if billed annually ($125 if billed monthly). Premium seats offer roughly 5× more usage limits and absence of certain usage caps relative to Standard seats.
As for API usage, the new GPT-Image-2.5 models are now available under GPT-Image-2.5 Flare and Sunburst, though OpenAI has not yet published separate pricing for these successors. Users rely on existing pricing for GPT-Image-2.0, which charges based on tokens: input text tokens, image input tokens (when editing), and output image tokens. For GPT-Image-2.0, input prompts are priced at $5 per million text tokens, image input at $8 per million tokens, and output image tokens at $30 per million tokens. Cached input rates are lower.
Who is it for?
Marketing teams needing assets like social media graphics, product photography, flyers, or posters. The improved control over text rendering, layout, and asset consistency makes it viable for campaign work.
Content creators and small businesses that want to generate or edit imagery in‐house, including modifying logos, product images, or promotional materials, with minimal back‐and‐forth with designers.
Design or brand operations teams looking to rapidly iterate visuals, especially with reference photos or sketches, and maintain coherence of style and subject matter across multiple versions.
Developers integrating image generation via API—for example, to build creative tools, software that personalizes visual content, or automation workflows involving graphic creation or photo editing.
When evaluating whether to adopt it, decision‐makers should consider volume of image usage, quality needs, and control over final appearance: businesses with frequent visual content demands will likely benefit most from the speed improvements and precision editing tools, while occasional users may be satisfied with lower usage plans.
Final thoughts
ChatGPT Images 2.5 represents a meaningful evolution in AI‐assisted visual design, offering better speed, sharper visual fidelity, more flexible editing, and greater control over the creative process. Its rollout across all user tiers and introduction of API variants makes it accessible for a variety of business contexts. While pricing for the new 2.5 models in the API is not yet published, the existing token‐based pricing under GPT-Image-2.0 serves as a useful benchmark. For organizations with ongoing content production needs, Images 2.5 may reduce time and design overhead; for those with occasional use, it still provides powerful capabilities, though cost‐efficiency will depend on usage scale and quality settings.
Visit the official website for more.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.