ChatGPT Images 2.5 :
فكّر، صمّم، وعدّل صوراً احترافية داخل AI STUDIOS

ChatGPT Images 2.5 (API name: GPT-Image-2.5) is OpenAI's latest image generation and editing model, released September 8, 2026. It builds on native reasoning ("Thinking") capabilities with sharper detail, more natural lighting, and up to 50% faster generation, and inside AI Studios it helps you create professional visuals up to 2K resolution with precise editing and multilingual text rendering.

ChatGPT Images 2.5 delivers sharper details, more natural lighting, and richer textures, with up to 50% lower generation latency than 2.0. It better preserves subjects from reference photos, follows editing instructions more reliably across multiple rounds, and handles transparent backgrounds more accurately. Resolution (up to 2K), aspect ratio support, and multilingual text rendering carry over unchanged from 2.0.

Open the Image Generator in AI Studios, select ChatGPT Images 2.5, add a prompt (and optional reference image), then generate, refine, and download your result. For complex briefs, enable Thinking mode to let the model plan and reason through layout before generating.

You can change specific elements while preserving composition, adjust colors and lighting, swap backgrounds, modify objects, transform styles, and create product variations — all with text instructions and pixel-level control. The model is particularly strong at preserving fine details, iconography, UI elements, and small text.

AI Studios uses OpenAI's ChatGPT Images 2.5 (GPT-Image-2.5) technology to power image generation and editing workflows. Feature availability and policies may vary by product environment.

Commercial use is allowed under AI Studios' terms and licensing. Please check the latest policy for your specific project or region.

AI Studios supports common formats like JPG and PNG for input, and lets you export high-quality images in multiple aspect ratios up to 2K resolution.

It's ideal for marketers, e-commerce teams, designers, agencies, and global brands who need high-fidelity, production-ready visuals with reasoning-driven accuracy, multilingual text rendering, and precise editing control.