AppSim-Bench - Chinese Apps: leaderboard
AppSim-Bench task-weighted accuracy over the Chinese app group. Higher is better. Source: arxiv.org/abs/2609.07712.
Metric: Task-weighted Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 60.07 | #51 |
| 2 | Gemini 3 Pro | 60.07 | #82 |
| 3 | Gemini 3.1 Pro (Preview) | 59.36 | #64 |
| 4 | GPT-5.5 | 58.66 | #33 |
| 5 | Seed-1.6 | 50.88 | #272 |
| 6 | Qwen 3.6 Plus | 39.58 | #124 |
| 7 | Qwen 3.6 Flash | 38.87 | #221 |
| 8 | GPT-5 | 37.81 | #105 |
| 9 | GPT-5.4 | 34.63 | #83 |
| 10 | Seed 1.8 | 34.63 | #143 |
| 11 | Claude Sonnet 4.6 | 31.1 | #94 |
| 12 | Claude Sonnet 4.5 | 24.03 | #148 |
| 13 | Claude Haiku 4.5 | 22.61 | #287 |
| 14 | Gemini 2.5 Pro | 21.2 | #159 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=appsim-bench-chinese-apps · How It Works · Data refreshed daily, snapshot 2026-11-05.