HoWToBench - Open Writing: leaderboard
HoWToBench (2026) tests long-form Chinese writing against curated human references; GPT-4o rates content traits 1 to 10 and format traits 0, 5 or 10, and the Tree-of-Writing method combines them with explicit per-instruction weights. Mean score on open-ended writing tasks with only an instruction; higher is better.
Metric: Tree-of-Writing score: GPT-4o rates each content trait 1-10 against the human reference and format traits 0, 5 or 10, combined with per-instruction LLM-assigned weights that sum to 1, mean over the 467 open-ended writing tasks of HoWToBench; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | HoWToBench DeepSeek-R1 (checkpoint unspecified) | 6.06 | |
| 2 | O3 Mini (2025-01-31) | 5.69 | #496 |
| 3 | Claude 3.5 Sonnet (20241022) | 5.43 | #648 |
| 4 | GPT-4o (2024-11-20) | 5.36 | #859 |
| 5 | Gemini 2.0 Flash | 5.33 | #723 |
| 6 | HoWToBench DeepSeek-V3 (checkpoint unspecified) | 5.31 | #666 |
| 7 | GLM-4-Plus-250111 | 5.28 | #831 |
| 8 | Doubao-pro-241225 | 5.14 | #810 |
| 9 | Claude 3 Haiku (20240307) | 4.85 | #2396 |
| 10 | Llama 3.3 70B Instruct | 4.47 | #1385 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=howtobench-open-writing · How It Works · Data refreshed daily, snapshot 2026-11-21.