Among editing actions, FLUX 3 Image Klein and FLUX 3 Image both improve most in Composition & Framing over their FLUX.2 predecessors

Our Image Arena taxonomy measures 7 editing actions and 10 use cases for Image Editing. Editing actions measure what an edit asks the model to do.

Composition & Framing is the biggest editing-action gain for both models: FLUX 3 Image jumps from the bottom third of the board into the top ten, and Klein into the top quarter. Reasoning-Based Edit is the next-biggest gain for both. Identity Preserving Edit remains a relative weak spot for both models.

➤ Composition & Framing: cropping, extending, reframing, and changes to camera view or perspective

➤ Reasoning-Based Edit: edits that need inference about the scene rather than a literal instruction

➤ Identity Preserving Edit: changes to people, faces and characters that must keep their identity intact