Introducing Images 2.6

Commenters focused on Images 2.6’s new animated-GIF and SVG capabilities. Some welcomed the ability to animate existing images and saw SVG generation as interesting, while others said GIFs exposed flicker, artifacts, and poor consistency between frames; one commenter argued sprite-sheet animation was an established technique. SVG output appears to involve converting a generated reference image into vectors, and a tester said API use requires specifying a separate language model. Reactions to the sample artwork were sharply divided: several called the stained-glass whale ugly or technically incoherent, while another thought it appealing. Some also objected to visible watermark artifacts, and a phone user criticized the sharing flow for sending recipients to ChatGPT rather than sharing a GIF directly.

Image quality comparisons produced another split. Several users said 2.6 performed worse than 2.5 or even 2.0 on their prompts and workflows, including design, fine detail, and text; one commenter cited a leaderboard result favoring 2.5. Others reported successes on difficult prompts or were impressed that the model could generate GIFs at all. Commenters also questioned the pricing: although token rates were unchanged, they calculated that low- and medium-quality 2.6 images use several times as many tokens, making each image more expensive. OpenAI’s claim of faster generation drew attention alongside those cost figures. Beyond product performance, one commenter saw generative imagery as a loss of human value, while another argued that many users accept the recognizable “AI look.”