UGI - Writing: leaderboard
The Writing column of the Uncensored General Intelligence leaderboard, a composite that weighs a model's intelligence together with measured properties of its prose: style, how much it repeats itself, and whether it keeps to the requested length. Target values for the metrics in the formula were estimated from human feedback so the score tracks the average person's preferences. Higher is better on a 0 to 100 scale.
Metric: Writing Score. Source: huggingface.co. Saturation forecast: Around January 2027. 1272 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash (Minimal) | 72.47 | #60 (Gemini 3.5 Flash) |
| 2 | Claude Opus 4.6 (Adaptive Reasoning, Max Effort) | 70.7 | #63 (Claude Opus 4.6) |
| 3 | Claude Opus 4.5 (20251101) (Thinking) | 70.34 | #74 (Claude Opus 4.5 (20251101)) |
| 4 | Gemini 3.5 Flash (High) | 69.71 | #60 (Gemini 3.5 Flash) |
| 5 | DeepSeek V4 Pro (Reasoning) | 68.4 | #102 (DeepSeek V4 Pro) |
| 6 | Claude Opus 4 (Non-reasoning) | 68.24 | #163 (Claude Opus 4) |
| 7 | Gemini 3.1 Pro (Preview) (High) | 68.22 | #58 (Gemini 3.1 Pro (Preview)) |
| 8 | Claude 3.7 Sonnet (20250219) (Thinking) | 68.01 | #201 (Claude 3.7 Sonnet (20250219)) |
| 9 | Gemini 3 Flash (Preview) (Medium) | 67.71 | #82 (Gemini 3 Flash (Preview)) |
| 10 | GPT-4.1 | 67.48 | #248 |
| 11 | GPT-4o ChatGPT | 67.37 | |
| 12 | Grok 4 | 67 | #181 |
| 13 | Claude Sonnet 4.5 (Non-reasoning) | 66.88 | #142 (Claude Sonnet 4.5) |
| 14 | O3 (2025-04-16) (Medium) | 66.83 | #121 (O3 (2025-04-16)) |
| 15 | Claude Opus 4 (20250514) (Thinking) | 66.83 | #154 (Claude Opus 4 (20250514)) |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=ugi-writing · How It Works · Data refreshed daily, snapshot 2026-10-20.