WritingBench: leaderboard
Generative writing benchmark covering writing quality, style control, factuality, and instruction following across writing tasks.
Metric: Score (%). Source: llm-stats.com. Status: saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 235B A22B 2507 (Thinking) | 88.3 |
| 2 | Qwen 3 Next 80B A3B Instruct | 87.3 |
| 3 | Qwen 3 VL 235B A22B (Thinking) | 86.7 |
| 4 | Qwen 3 VL 32B (Thinking) | 86.2 |
| 5 | Qwen 3 VL 235B A22B Instruct | 85.5 |
| 6 | Qwen 3 VL 8B (Thinking) | 85.5 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 85.2 |
| 8 | Qwen 3 VL 30B A3B (Thinking) | 85.2 |
| 9 | Qwen 3 Next 80B A3B (Thinking) | 84.6 |
| 10 | Qwen 3 VL 4B (Thinking) | 84 |
| 11 | Qwen 3 VL 8B Instruct | 83.1 |
| 12 | Qwen 3 VL 32B Instruct | 82.9 |
| 13 | Qwen 3 VL 30B A3B Instruct | 82.6 |
| 14 | Qwen 3 VL 4B Instruct | 82.5 |
Interactive version: theaggregate.ai/benchmark?slug=writingbench · How It Works · Data refreshed daily, snapshot 2026-09-05.