WeEdit (Multilingual) - Translate: leaderboard
Metric: Mean instruction-adherence score (0-10 judge scale) on the translate operation of the WeEdit Multilingual Benchmark (2,000 text-centric image-editing cases), rated by a Gemini-3-Pro judge that writes a chain-of-thought rationale before scoring whether the edited image fulfils every requirement of the instruction; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Nano Banana Pro | 6.34 | |
| 2 | Seedream 4.5 | 3.04 | |
| 3 | GPT-Image-1.5 | 2.97 | |
| 4 | Gemini 2.5 Flash Image | 1.83 | |
| 5 | HunyuanImage-3.0-Instruct | 1.75 | |
| 6 | FireRed-Image-Edit | 1.62 | |
| 7 | Qwen-Image-Edit-2511 | 1.56 | |
| 8 | EMU3.5 | 1.47 | |
| 9 | Qwen-Image-Edit-2509 | 1.44 | |
| 10 | FLUX.2-dev | 1.4 |
Interactive version: theaggregate.ai/benchmark?slug=weedit-multilingual-translate · How It Works · Data refreshed daily, snapshot 2026-10-11.