Benchmarks / 3D
Slop Score
Leaderboard and analysis tool scoring how much model writing resembles AI slop via overused words, contrast patterns, and trigrams. Maintained by Sam Paech. Score: Weighted composite of slop-word frequency (60%), 'not-x-but-y' contrast patterns (25%), and slop trigram frequency (15%) on outputs from a standardized prompt set; lower is better.
Slop Score
Lower is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Kimi K2.0905 | – | 18 |
| 2 | Sonnet 4.5 | high | 19 |
| 3 | Gemma 3 27B IT | medium | 21 |
| 4 | Kimi K2 | medium | 24 |
| 5 | GPT-5 Mini | high | 26 |
| 6 | Sonnet 4 | high | 27 |
| 7 | o3 | high | 32 |
| 8 | GPT-5 Nano | high | 34 |
| 9 | Llama 4 Maverick | medium | 49 |
| 10 | Mistral Small 3.2 | none | 53 |
| 11 | DeepSeek V3.2 | high | 54 |
| 12 | Mistral | none | 54 |
| 13 | GLM 4.5 | xhigh | 60 |
| 14 | Qwen 3 4B | none | 62 |
| 15 | DeepSeek R1 | xhigh | 68 |
| 16 | Gemma 3 12B IT | – | 71 |
| 17 | Gemini 2.5 Flash | medium | 78 |
| 18 | Gemma 3 4B IT | – | 85 |