SuiChat-CN - Without Keywords: leaderboard
Metric: Macro F1 (%) of suicide-risk level classification of Chinese group-chat topics into six levels (L0 to L5) on the SuiChat-CN test set, macro F1 over the levels, with the risk keywords removed from the input; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 | 76.3 |
| 2 | GLM-5.1 | 75.1 |
| 3 | Qwen 3.5 Plus | 75.1 |
| 4 | Kimi K2 | 74.7 |
| 5 | Gemini 2.5 Pro | 74.5 |
| 6 | Seed-OSS-36B-Instruct | 74.2 |
| 7 | Qwen Max | 73.1 |
| 8 | Qwen 3.5 397B A17B | 72.9 |
| 9 | GLM-4 32B (0414) | 71.6 |
| 10 | Qwen 3 235B A22B | 70.5 |
| 11 | GPT-5 | 70.1 |
| 12 | GLM-5 | 69.9 |
| 13 | Llama 3.3 70B Instruct | 69.8 |
| 14 | Kimi K2.5 | 69.3 |
| 15 | GLM-4.6 | 69 |
Interactive version: theaggregate.ai/benchmark?slug=suichat-cn-without-keywords · How It Works · Data refreshed daily, snapshot 2026-10-07.