SEAL Showdown: leaderboard
Scale AI human expert evaluation of LLMs across real-world tasks. Arena-style scoring based on pairwise comparisons by trained human raters.
Metric: Arena Score. Source: scale.com. Status: saturation imminent. 47 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 1255.4 |
| 2 | Gemini 3 Flash | 1206.5 |
| 3 | GPT-5.5 | 1116.4 |
| 4 | Claude Opus 4.7 (Thinking) | 1113.1 |
| 5 | Claude Opus 4.6 (Thinking) | 1109 |
| 6 | Claude Opus 4.1 (20250805) (Thinking) | 1093 |
| 7 | GPT-5.1 (Medium) | 1092.9 |
| 8 | Claude Opus 4.7 | 1087.1 |
| 9 | Claude Opus 4.6 | 1086.1 |
| 10 | Kimi K2.5 (Thinking) | 1085.1 |
| 11 | GPT-5 Chat | 1081.5 |
| 12 | DeepSeek V3.2 | 1077.7 |
| 13 | GPT-5.5 (Medium) | 1077.6 |
| 14 | Claude Sonnet 4.6 | 1072.4 |
| 15 | Claude Opus 4.5 (20251101) | 1066 |
Interactive version: theaggregate.ai/benchmark?slug=seal-showdown · How It Works · Data refreshed daily, snapshot 2026-09-05.