source
Artificial Analysis
Benchmarks · /benchmarks/artificial-analysis-image-edit
Image Editing
Artificial Analysis image arena leaderboard for instruction-based image editing.
Passport
Visible tradeoffsThis is a human preference signal, so it tells you what people liked side by side, not what is formally correct.
metric
Arena rating (rating)
judge
Human
direction
higher better
group id
aa_image_edit_current
domain
Image editing
What it measures vs what it misses
✓ Measures
Observed user preference over edited outputs from the same source image and edit request. How well a model preserves intent while making preferred visual changes.
✗ Misses
Localized edit fidelity metrics. Safety and production workflow guarantees.
Why this countsIt would matter for editing and instruction-following once verified public receipts exist in the catalog.Comparable-group ruleThis percentile only compares models inside the exact benchmark/version group shown here. It is not a universal score.What it missesThis slice is currently limited because the product does not yet carry first-class image-editing receipts.
Leaderboard · this benchmark version
#1 · GPT Image 1.5
AA · May 1, 2026
1,267
#2 · GPT Image 2 (high)
AA · May 1, 2026
1,252
#3 · Nano Banana Pro (Gemini 3 Pro Image)
AA · May 1, 2026
1,248
#4 · Nano Banana 2 (Gemini 3.1 Flash Image Preview)
AA · May 1, 2026
1,231
#5 · HunyuanImage 3.0 Instruct (Fal)
AA · May 1, 2026
1,229