LongBench Pro - Chinese - Thinking: leaderboard
Evaluates long-context understanding across English and Chinese tasks at four difficulty levels. Non-thinking and thinking results use separate published columns; missing settings remain absent. This series reports Chinese - Thinking using Score (%). Source: the original paper result table.
Metric: Score (%). Source: arxiv.org. Saturation forecast: Around February 2027. 46 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 74.49 | #159 |
| 2 | GPT-5 | 71.97 | #105 |
| 3 | Claude Sonnet 4 | 68.65 | #207 |
| 4 | DeepSeek V3.2 | 67.75 | #210 |
| 5 | Gemini 2.5 Flash | 67.59 | #255 |
| 6 | Qwen 3 235B A22B 2507 (Thinking) | 67.12 | #263 |
| 7 | DeepSeek V3.1 | 66.26 | #275 |
| 8 | Qwen 3 Next 80B A3B (Thinking) | 64.99 | #312 |
| 9 | DeepSeek R1 0528 | 63.89 | #232 |
| 10 | Qwen 3 235B A22B 2507 Instruct | 63.65 | #315 |
| 11 | Qwen 3 Next 80B A3B Instruct | 61.22 | #381 |
| 12 | GLM-4.6 | 59.92 | #265 |
| 13 | Qwen 3 30B A3B 2507 (Thinking) | 59.22 | #379 |
| 14 | Claude 3.7 Sonnet | 58.84 | #259 |
| 15 | DeepSeek R1 | 58.13 | #262 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=longbench-pro-chinese-thinking · How It Works · Data refreshed daily, snapshot 2026-11-05.