OpenAI announced ChatGPT Images 2.5 today, a new image model that promises sharper details, more natural lighting, richer textures and faster generation compared with Images 2.0. According to the company, users create more than 3 billion images per week across ChatGPT Images and the GPT‑Image API models, and Images 2.5 aims to make those creative workflows both faster and more reliable.
Key improvements in Images 2.5
- Better reference fidelity: the model preserves characteristics of subjects in reference photos more effectively, making people, places and objects more recognizable when placed in new settings or visual styles.
- More natural lighting and texture: generated images have more realistic illumination and surface detail.
- More precise editing across multiple steps: Images 2.5 follows editing instructions more reliably across several edits, so earlier changes remain consistent and image quality does not degrade over time.
- Faster generation: OpenAI reports up to a 50% reduction in generation latency compared with Images 2.0.
New ChatGPT features
The update also introduces several product features inside ChatGPT:
- Sketch: a new tool that lets users draw directly in ChatGPT as a visual guide; the system can turn rough sketches into finished images when combined with a style or detail description. To use it, type “@Sketch” in ChatGPT.
- Templates: prebuilt templates for common formats (for example, poster or product photography) to help users start from a layout rather than a blank canvas.
- Image comments: users can place comments directly on images to support more focused editing.
- Shareable prompts: when sharing an image, users can now include the prompt used to generate it so others can try the same idea with their own photos and details.
API updates for developers
OpenAI is releasing two new models in the API:
- GPT‑Image‑2.5 Flare: brings the same quality, editing and speed improvements to the API and is intended as the default choice for many applications. Flare delivers higher-quality images than GPT‑Image‑2 with 50% lower latency.
- GPT‑Image‑2.5 Sunburst: designed for premium visual workflows that require tighter control across edits and may involve longer generation times; suited for production-ready campaign creative and polished product imagery.
OpenAI positions Flare for a wide range of use cases — creator and social content, product experiences, visual search, rapid image prototyping, and high-volume generation — while Sunburst targets workflows that demand precise control and consistency.
Consistency and handling of complex instructions
Images 2.5 is better at understanding complex visual instructions and translating them into coherent results. It handles real-world information more accurately, supports more complex layouts including transparent backgrounds, and reflects visual styles more faithfully. For teams and businesses this means creative briefs that are more dependable: the model is less likely to drift away from a requested visual direction, composition or detail as instructions become more specific.
For API-driven, reference-led workflows, this fidelity enables targeted updates — for example changing a single product or background element — while preserving subject, composition and brand treatment around it.
Safety and provenance
OpenAI says Images 2.5 builds on its existing safeguards, performing checks on prompts and images to help prevent harmful outputs. The company continues to use C2PA metadata and invisible watermarking to help identify images created with its tools. Further evaluation and implementation details are documented in the system card.
Availability
Images 2.5 is rolling out today to ChatGPT, ChatGPT Work and Codex users across desktop, mobile and web, across all tiers. GPT‑Image‑2.5 Flare and GPT‑Image‑2.5 Sunburst are available via the API; pricing details are provided in OpenAI’s documentation.
In short, OpenAI positions Images 2.5 as an upgrade that delivers improved detail, speed and multi-step editing reliability, together with new user features (Sketch, templates, on-image comments) and API models aimed at both high-volume and premium visual workflows.
