Benchmarks · creative
LLM Creative Story-Writing Benchmark
Pairwise comparison of short stories written to the same constrained creative briefs, with LLM evaluator choices combined into a relative comparison score. Maintained by Lech Mazur. Score: Relative comparison score from pairwise LLM evaluator judgments (Thurstone-style rating; zero is near the middle of the comparison set).
52 models · 57 results · data as of 2026-10-29 · raw.githubusercontent.com ↗
Leaderboard
Fable 5.1 high
All results
57 rows| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Fable 5.1 | high | 3.7 |
| 2 | Opus 5.5 | high | 3.7 |
| 3 | Opus 5 | xhigh | 3.7 |
| 4 | GPT-6 Astra | high | 3.4 |
| 5 | Opus 5 | high | 3.3 |
| 6 | GLM 5.3 | max | 3.0 |
| 7 | Fable 5 | high | 2.8 |
| 8 | Kimi K3.1 | high | 2.7 |
| 9 | GPT-5.6 Sol | xhigh | 2.4 |
| 10 | GPT-5.5 | xhigh | 2.4 |
| 11 | Kimi K3 | max | 2.3 |
| 12 | GPT-5.6 Sol | high | 2.2 |
| 13 | GPT-5.4 | medium | 1.9 |
| 14 | GPT-5.4 | xhigh | 1.9 |
| 15 | Opus 4.7 | high | 1.7 |
| 16 | Sonnet 4.6 | max | 1.6 |
| 17 | MiMo V2.6 Pro | high | 1.5 |
| 18 | Opus 4.6 | max | 1.1 |
| 19 | Gemini 4 Argon | high | 0.7 |
| 20 | Opus 4.8 | xhigh | 0.7 |
| 21 | Muse Spark 1.1 | high | 0.7 |
| 22 | Muse Spark 1.3 | high | 0.5 |
| 23 | DeepSeek V4 Pro | high | 0.4 |
| 24 | GLM 5.2 | max | 0.4 |
| 25 | Grok 4.7 | high | 0.4 |
| 26 | GPT-5.2 | medium | 0.3 |
| 27 | Opus 4.8 | high | 0.2 |
| 28 | Qwen 3.8 Max | max | 0.1 |
| 29 | Gemini 3.8 Flash | high | 0.1 |
| 30 | Kimi K2.6 | max | 0.0 |