SuperCLUE-Faith - Overall: leaderboard
SuperCLUE is the Chinese-language evaluation programme run by CLUE; its specialised boards each cover one professional domain and are graded by the organisers rather than self-reported. This board covers faithfulness, meaning whether answers stay grounded in the supplied material.
Metric: Score. Source: www.superclueai.com. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (High) | 93.19 | #110 (GPT-5.2) |
| 2 | Gemini 3 Pro (Preview) | 90.96 | #66 |
| 3 | Doubao-Seed-1.6-251015 (Thinking) | 86.2 | #207 (Doubao-Seed-1.6-251015) |
| 4 | GLM-4.7 | 85.11 | #192 |
| 5 | Tencent HY 2.0 Think | 81.83 | #198 |
| 6 | DeepSeek V3.2 (Thinking) | 79.51 | #205 (DeepSeek V3.2) |
| 7 | Kimi K2 (Thinking) | 78.99 | #243 (Kimi K2) |
| 8 | MiniMax-M2.1 | 73.29 | #289 |
| 9 | Qwen 3 235B A22B 2507 (Thinking) | 71.03 | #260 (Qwen 3 235B A22B 2507) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=superclue-faith-overall · How It Works · Data refreshed daily, snapshot 2026-10-20.