Chatbot Arena ELO rating for creative writing prompts — how humans rate model responses to storytelling, poetry, copywriting, and narrative tasks in blind pairwise tests.
Human judges evaluate creative quality holistically; blind pairwise setup reduces bias; large vote count from diverse judges.
Creative quality is subjective; voters may favor fluent prose over originality; English-centric evaluation. Data: Arena (lmarena.ai) leaderboard dataset, https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset, licensed CC-BY-4.0; rows are converted to the ELO fields shown here.
Higher is better. ELO-like rating for creative writing tasks.
Creative quality is subjective.
| # | Model | Score |
|---|---|---|
| 1 | 1494.0 | |
| 2 | 1490.0 | |
| 3 | 1485.0 | |
| 4 | 1472.0 | |
| 5 | 1468.0 | |
| 6 | 1467.0 | |
| 7 | 1462.0 | |
| 8 | 1459.0 | |
| 9 | 1454.0 | |
| 10 | 1453.0 | |
| 11 | 1452.0 | |
| 12 | 1449.0 | |
| 13 | 1447.0 | |
| 14 | 1447.0 | |
| 15 | 1444.0 | |
| 16 | 1444.0 | |
| 17 | 1444.0 | |
| 18 | 1442.0 | |
| 19 | 1435.0 | |
| 20 | 1432.0 | |
| 21 | 1431.0 | |
| 22 | 1430.0 | |
| 23 | 1430.0 | |
| 24 | 1428.0 | |
| 25 | 1424.0 | |
| 26 | 1423.0 | |
| 27 | 1423.0 | |
| 28 | 1422.0 | |
| 29 | 1420.0 | |
| 30 | 1415.0 | |
| 31 | 1412.0 | |
| 32 | 1411.0 | |
| 33 | 1410.0 | |
| 34 | 1409.0 | |
| 35 | 1406.0 | |
| 36 | 1405.0 | |
| 37 | 1404.0 | |
| 38 | 1404.0 | |
| 39 | 1401.0 | |
| 40 | 1400.0 | |
| 41 | 1399.0 | |
| 42 | 1399.0 | |
| 43 | 1397.0 | |
| 44 | 1397.0 | |
| 45 | 1395.0 | |
| 46 | 1395.0 | |
| 47 | 1395.0 | |
| 48 | 1395.0 | |
| 49 | 1394.0 | |
| 50 | 1394.0 |