P-Bench - Qwen-Image-Bench: leaderboard
P-Bench scores text-to-image providers over the Qwen-Image-Bench prompt set with P-Judge, the overall quality judge PrunaAI runs itself, alongside price per image and median generation time.
Metric: P-Judge Overall. Source: huggingface.co. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT Image 2 | 59.02 |
| 2 | P Image 2 Ideogram Very High 1K | 58.28 |
| 3 | GPT Image 1.5 | 57.86 |
| 4 | Nano Banana Pro | 56.45 |
| 5 | Nano Banana 2.0 | 56.36 |
| 6 | Qwen Image 2.0 Pro | 56.18 |
| 7 | P Image 2 Ideogram High 1K | 55.75 |
| 8 | Seedream 5.0 | 55.72 |
| 9 | Seedream 4.5 | 55.66 |
| 10 | Seedream 4.0 | 55.24 |
Interactive version: theaggregate.ai/benchmark?slug=p-bench-qwen-image-bench · How It Works · Data refreshed daily, snapshot 2026-09-05.