OpenAI Unveils ChatGPT Images 2.5 with Faster Generation and More Precise Editing

OpenAI has introduced ChatGPT Images 2.5, the latest upgrade to its text-to-image service. The new version is available immediately to all ChatGPT, ChatGPT Work, and Codex users on desktop, mobile, and web, spanning free-tier and paid subscriptions.

9to5Mac

OpenAI reports a reduction in generation latency of up to 50 % compared with Images 2.0, and internal tests show the Flare model delivering speeds two to four times faster than its predecessor. The update also improves visual fidelity: images exhibit more natural lighting, richer textures, and stronger preservation of subjects from reference photos. Editing reliability across multiple turns has been heightened, allowing users to modify specific areas without altering the surrounding composition. OpenAI announced that combined usage of the ChatGPT interface and the GPT-Image API now reaches roughly 3 billion images per week (about 430 million per day). New user-facing tools include a “Sketch” panel for hand-drawn inputs, template presets for common formats such as posters and product shots, and prompt-sharing functionality that lets users distribute complete generation prompts.

OpenAI Launches GPT-Image-2.5 with Faster Generation and Enhanced Editing 9to5Mac

Developers can choose between two API endpoints launched alongside Images 2.5. GPT-Image-2.5 Flare is positioned as the default model, emphasizing speed and batch generation for high-volume scenarios like social content and rapid prototyping. GPT-Image-2.5 Sunburst targets premium visual workflows, offering tighter edit control for production-grade marketing assets and polished product imagery. OpenAI describes Flare as delivering higher-quality images at 50 % lower latency, while Sunburst trades longer generation times for finer precision.

9to5Mac

Adobe has confirmed that the Images 2.5 model is being integrated into its Firefly platform. Matt Chotin, senior director of Adobe Products, notes that version 2.5 provides faster generation speeds and more consistent resolution, keeping images sharp and realistic even after multiple refinement cycles. The introduction of speed- and quality-tiered API models mirrors OpenAI’s recent language-model pricing structure and signals a broader push to align multimodal capabilities with upcoming large-scale models such as GPT-7.

OpenAI Launches GPT-Image-2.5 with Faster Generation and Enhanced Editing 9to5Mac

• “Images 2.5 produces more natural lighting and richer textures, is better at preserving the subjects in your reference photos, and follows editing instructions more reliably across multiple turns,” — Plus OpenAI

9to5Mac