SEAL - Chinese: leaderboard
1,000 single-turn Chinese-language prompts across writing, coding, daily life, travel, and creative tasks, measuring helpfulness, fluency, and cultural appropriateness in Chinese.
Metric: Score. Source: scale.com. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 (2024-12-17) | 1165 | #149 |
| 2 | O3 Mini | 1156 | #272 |
| 3 | O1 Preview | 1120 | #193 |
| 4 | Gemini 2.0 Flash | 1115 | #342 |
| 5 | Gemini 1.5 Pro | 1077 | #445 |
| 6 | Gemini 2.0 Flash (Thinking) | 1060 | #342 (Gemini 2.0 Flash) |
| 7 | DeepSeek R1 | 1052 | #251 |
| 8 | DeepSeek V3 | 1031 | #321 |
| 9 | GPT-4o (2024-08-06) | 1029 | #336 |
| 10 | Gemini 1.5 Flash | 1015 | #618 |
| 11 | Mistral Large 2 (Jul) | 1006 | #484 |
| 12 | DeepSeek V2 Chat | 996 | #563 |
| 13 | GPT-4 | 985 | #438 |
| 14 | aya-expanse-32B | 967 | #692 |
| 15 | Gemma 2 27B | 966 | #806 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=seal-chinese · How It Works · Data refreshed daily, snapshot 2026-10-20.