ChatGPT Images 2.5: Faster, more precise, but not the same for everyone
decoder ·
OpenAI has launched ChatGPT Images 2.5, introducing two new image generation models: GPT-Image-2.5 Flare for faster, general use and GPT-Image-2.5 Sunburst for more precise editing tasks. The company claims sharper details, improved subject consistency, and up to 50% lower latency compared to the previous version. New features in ChatGPT include a Sketch drawing tool, ready-made templates, and shareable prompts. The models currently hold the top two spots on the Arena text-to-image leaderboard, though the ratings are preliminary. OpenAI uses the C2PA standard and Google DeepMind's SynthID for provenance labeling. The rollout covers all ChatGPT users, Codex, and the API, but there is no explicit control in ChatGPT to choose between the two models. Tests suggest the editing improvements work reliably in ChatGPT Work mode, while the standard Chat mode appears to use the weaker model. Pricing varies by quality tier, with a new 'max' tier costing roughly $0.21 for a 1024x1024 image.
decoder