Benchmark
Arena ELO Image Edit
People compare model answers and choose which they prefer. Higher scores are better.
All human preference benchmarksScores from LMArena data.
- Models measured
- 58
- Scores recorded
- 64
- Sources
- 1
- IQ weight
- 1.04
Score distribution
- 1,297
- Mean
- 1,302.1
- Median
- 113.8
- Std dev
- 1,150.2
- 10th percentile
- 1,437.4
- 90th percentile
- 996.1
- Min
This benchmark has a weight of 1.04 for image-generation models, measured over 46 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 51 | Doubao Seedream 4-0fal· LMArena data | 1,209.4 |
| 52 | Reve v1.1 Fast· LMArena data | 1,206.7 |
| 53 | Reve Edit Fast· LMArena data | 1,198 |
| 54 | FLUX.2 [klein] 4B· LMArena data | 1,187.1 |
| 55 | Flux Kontext Max· LMArena data | 1,182.1 |
| 56 | Wan 2.5 Preview· LMArena data | 1,179.7 |
| 57 | Flux Kontext Pro· LMArena data | 1,177.3 |
| 58 | Flux Kontext Dev· LMArena data | 1,150.2 |
| 59 | SeedEdit 3.0· LMArena data | 1,139.1 |
| 60 | GPT Image 1· LMArena data | 1,138.8 |
| 61 | GPT-Image-1 mini· LMArena data | 1,124 |
| 62 | Gemini 2.0 Flash Image Generation· LMArena data | 1,081.9 |
| 63 | BAGEL· LMArena data | 1,026.6 |
| 64 | Step1X-Edit· LMArena data | 996.1 |
Showing 51–64 of 64 results.
Data credits: LMArena, CC BY 4.0 (licence)