Chatbot Arena (Text - Korean): leaderboard
The arena.ai text board restricted to battles whose prompt the site tags as Korean. Each battle is one user prompt answered by two anonymous models, with the voter choosing the better reply. The Bradley-Terry rating is refit over this subset of votes alone, so the order can differ from the overall text table and the confidence intervals are wider, and higher is better.
Metric: Arena Score. Source: arena.ai. 281 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | 1514 | #9 (Claude Fable 5.1) |
| 2 | Claude Opus 5 (Max) | 1499 | #12 (Claude Opus 5) |
| 3 | Muse Spark 1.3 (Max) | 1497 | #24 (Muse Spark 1.3) |
| 4 | Claude Fable 5 (High) | 1497 | #16 (Claude Fable 5) |
| 5 | Claude Opus 5 (High) | 1495 | #12 (Claude Opus 5) |
| 6 | Gemini 3.7 Flash (High) | 1480 | #23 (Gemini 3.7 Flash) |
| 7 | GLM-5.3 (Max) | 1477 | #39 (GLM-5.3) |
| 8 | Muse Spark 1.1 | 1475 | #52 |
| 9 | Muse Spark | 1474 | #64 |
| 10 | Claude Opus 4.7 | 1471 | #49 |
| 11 | Kimi K3 (Max) | 1470 | #30 (Kimi K3) |
| 12 | DeepSeek V4.1 Flash (Max) | 1467 | #36 (DeepSeek V4.1 Flash) |
| 13 | Claude Opus 4.6 | 1464 | #63 |
| 14 | DeepSeek V4 Pro (0813) (High) | 1464 | #53 (DeepSeek V4 Pro (0813)) |
| 15 | Claude Opus 4.7 (High) | 1462 | #49 (Claude Opus 4.7) |
Interactive version: theaggregate.ai/benchmark?slug=chatbot-arena-text-korean · How It Works · Data refreshed daily, snapshot 2026-10-20.