LMArena Text Arena: leaderboard
Human voters on LMArena send one prompt to two anonymous models and pick the better reply, and a Bradley-Terry fit over the accumulated votes gives each model a single rating. This board takes those ratings from benchmarklist.com, which marks each row as imported, rather than reading LMArena directly, and is labelled self-reported on this site. Higher is better. Source: arena.ai/faq.
Metric: Arena rating (imported). Source: lmarena.ai. Source: benchmarklist.com. Saturation forecast: Not forecast. 31 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 | 1507.59 | #16 |
| 2 | Claude Opus 4.6 (High) | 1504.74 | #63 (Claude Opus 4.6) |
| 3 | Claude Opus 4.7 (High) | 1502.05 | #49 (Claude Opus 4.7) |
| 4 | Muse Spark 1.2 (xHigh) | 1498.64 | #62 (Muse Spark 1.2) |
| 5 | Claude Opus 4.6 | 1497.55 | #63 |
| 6 | Claude Opus 4.7 | 1494.36 | #49 |
| 7 | Gemini 3.8 Flash (High) | 1493.74 | #21 (Gemini 3.8 Flash) |
| 8 | Claude Opus 5 (High) | 1492.01 | #12 (Claude Opus 5) |
| 9 | Muse Spark 1.1 | 1491.45 | #52 |
| 10 | Gemini 3.7 Flash (High) | 1490.92 | #23 (Gemini 3.7 Flash) |
| 11 | Kimi K3 (Max) | 1488.94 | #30 (Kimi K3) |
| 12 | Muse Spark | 1488.25 | #64 |
| 13 | Claude Opus 5 (Max) | 1487.52 | #12 (Claude Opus 5) |
| 14 | GLM-5.3 (Max) | 1486.94 | #39 (GLM-5.3) |
| 15 | Gemini 3.1 Pro (Preview) | 1486.7 | #58 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=lmarena-text-arena · How It Works · Data refreshed daily, snapshot 2026-10-20.