OpenAI Releases ChatGPT Images 2.5… 50% Faster Generation and New API Models

OpenAI has released its latest state-of-the-art image model, ChatGPT Images 2.5. This model offers faster responses by reducing generation latency by up to 50% compared to previous versions, while delivering natural lighting and rich textures. Additionally, the reliability of subject preservation from reference photos and adherence to multi-turn editing instructions has significantly improved.

Key Features and Improvements

• Precision Editing: Modifies only the requested parts while maintaining other details (subject, composition, brand handling, etc.). It allows for single-element updates even in complex backgrounds.
• Multi-turn Consistency: Maintains previous changes during long conversations and applies new edits without quality degradation, making it suitable for production workflows.
• New User Features:
  • Sketch: Uses hand-drawn sketches as final image references to generate high-quality results.
  • Templates: Provides templates for popular formats such as flyers and product photos, allowing users to start quickly instead of beginning with a blank canvas.
  • Prompt Sharing: Allows sharing of the prompts used for image generation, enabling others to reproduce or modify the same style.

New API Models and Availability

Two new models are available via the API. GPT-Image-2.5 Flare is the base model, featuring improved speed, quality, and editing capabilities, making it suitable for bulk generation and prototyping. GPT-Image-2.5 Sunburst is designed for premium workflows, offering higher precision and editing control, though generation takes slightly longer. Partners such as Adobe and Higgsfield AI have already confirmed integration and performance improvements with these models.

ChatGPT Images 2.5 is being rolled out sequentially across all tiers on desktop, mobile, and web for ChatGPT, ChatGPT Work, and Codex users. The API offers immediate access to GPT-Image-2.5 Flare and Sunburst.