Chinese Social Pragmatic Inference - Comment Only: leaderboard
The same Chinese social pragmatic inference items shown the comment alone, without the interaction context that gives it its meaning; leave-writer-out accuracy. Higher is better. Source: arxiv.org/abs/2609.04384.
Metric: Leave-writer-out Accuracy (%). Source: arxiv.org. Saturation forecast: Not forecast. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 9B | 33.58 | #391 |
| 2 | Llama 4 Maverick | 32.09 | #486 |
| 3 | DeepSeek V4 Pro | 31.2 | #112 |
| 4 | Mistral Medium 3 | 30.49 | #507 |
| 5 | MiniMax-M2.5 | 29.69 | #316 |
| 6 | GPT-5.5 | 28.4 | #33 |
| 7 | Claude Opus 4.7 | 26.79 | #51 |
| 8 | Kimi K2.6 | 24.91 | #113 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=chinese-social-pragmatic-inference-comment-only · How It Works · Data refreshed daily, snapshot 2026-11-05.