WISRD Benchmark Tests AI Image Editing Models with Visual IQ Tests

HirokatuKataoka · x · 2026-08-05

WISRD: Worksheet Image-Space Rule Discovery

The paper introduces WISRD, a novel end-to-end visual reasoning benchmark. It challenges image-editing models to solve problems entirely in image space—reading visual instructions, discovering rules, and writing answers directly onto the image without leaving the visual modality.

Core Challenges

The models must demonstrate multiple capabilities:

Frontier Model Performance

Key Finding

Analysis reveals that current image-editing models can partially rely on rendered in-image instructions, even ignoring external text prompts.

Original post →

More from Multimodal

Multimodal channel →