LLM Creative Story Writing Win Probability
Chat & Writing · 2025-01-05
Secondary metric from Lech Mazur's Creative Story-Writing benchmark, reporting each model's estimated win probability against the evaluated pool as a 0-1 probability.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 0.9 |
| GPT-5.5 | 0.9 |
| GPT-5.6 Sol | 0.9 |
| GPT-5.4 | 0.9 |
| Opus 4.7 | 0.8 |
| Sonnet 4.6 | 0.8 |
| Opus 4.6 | 0.8 |
| Opus 4.8 | 0.7 |
| Muse Spark 1.1 | 0.7 |
| GPT-5.2 | 0.7 |
| GLM-5.2 | 0.6 |
| Kimi K2.6 | 0.6 |
| MiniMax M3 | 0.6 |
| Mistral Medium 3.1 | 0.5 |
| DeepSeek-V4-Pro | 0.5 |
| MiMo-V2.5-Pro | 0.5 |
| Qwen3 Max (rolling alias) | 0.5 |
| Qwen3.6-Max-Preview | 0.5 |
| GLM-5.1 | 0.4 |
| Kimi K2.5 | 0.4 |
| ERNIE 5.1 | 0.4 |
| MiMo-V2-Pro | 0.4 |
| Mistral Large 3 675B Instruct 2512 | 0.3 |
| Gemma 4 31B IT | 0.3 |
| Gemini 3.5 Flash | 0.3 |
| Doubao Seed 2.0 Pro | 0.3 |
| Gemini 3.1 Pro Preview | 0.3 |
| Qwen3.6 Plus (2026-04-02) | 0.3 |
| Mistral Medium 3.5 | 0.2 |
| Qwen3.7-Max | 0.2 |
| DeepSeek-V3.2 | 0.2 |
| gpt-oss-120b | 0.1 |
| MiniMax M2.7 | 0.1 |
| Grok 4.3 | 0.1 |
| Grok 4.5 | 0.0 |