Arena-T2I Hard - Commercial: leaderboard

Metric: Faithfulness on the Arena-T2I Hard prompts whose primary category is commercial: dependency-aware faithfulness yes-ratio (%; share of the about 30 decomposed yes/no constraint questions per prompt that a gemini-3-flash judge answers yes for the generated image, a question counted only when its parent questions pass); higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.

Top models

#ModelScore
1Grok Imagine Image 2026030685.5
2gemini-3-pro-image-preview-2k84.2
3recraft-v481.2
4gemini-2.5-flash-image75.4
5gpt-image-1.5-high-fidelity75.3
6wan2.6-t2i-v271.5
7imagen-4.0-generate-00166.1
8gpt-image-165.8
9imagen-4.0-ultra-generate-00157.7
10hunyuan-image-3.0-fal51

Interactive version: theaggregate.ai/benchmark?slug=arena-t2i-hard-commercial · How It Works · Data refreshed daily, snapshot 2026-09-29.