Benchmarks · creative
EQ-Bench Longform Creative Writing
LLM-judged benchmark in which models plan and write a short story/novella over 8x 1000-word turns, scored 0-100 across 14 rubric dimensions. Maintained by EQ-Bench. Score: Score (0-100): average of all chapter scores plus the final scored piece across 14 rubric dimensions, with the forced poetry/metaphor criterion weighted 5x and a long-context degradation penalty applied.
105 models · 105 results · data as of 2026-10-29 · eqbench.com ↗
Leaderboard
Opus 5 max
All results
105 rows| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 86.3% |
| 2 | Fable 5.1 | max | 85.3% |
| 3 | Opus 5.5 | max | 84.6% |
| 4 | Fable 5 | max | 83.0% |
| 5 | Grok 4.7 | max | 83.0% |
| 6 | GPT-6 Astra | max | 82.8% |
| 7 | GPT-6 Sol | max | 82.8% |
| 8 | Muse Spark 1.3 | max | 82.8% |
| 9 | Opus 4.7 | max | 81.8% |
| 10 | GLM 5.3 | max | 81.8% |
| 11 | GPT-5.6 Sol | max | 81.7% |
| 12 | Muse Spark 1.2 | max | 81.5% |
| 13 | Gemini 4 Argon | max | 80.9% |
| 14 | Opus 4.8 | max | 80.8% |
| 15 | Sonnet 4.6 | max | 79.9% |
| 16 | Kimi K3 | max | 79.6% |
| 17 | Kimi K2.6 | max | 78.5% |
| 18 | Sonnet 5 | max | 78.3% |
| 19 | GPT-5.4 | xhigh | 78.3% |
| 20 | GPT-5.5 | xhigh | 78.2% |
| 21 | GPT-5.6 Terra | max | 78.0% |
| 22 | GLM 5.2 | max | 77.9% |
| 23 | Opus 4.6 | max | 77.7% |
| 24 | Gemini 3.8 Flash | max | 76.8% |
| 25 | DeepSeek V4 Pro | max | 75.6% |
| 26 | GPT-5.6 Luna | max | 75.5% |
| 27 | Kimi K2.5 | max | 74.9% |
| 28 | DeepSeek V4.1 Flash | max | 74.0% |
| 29 | MiMo V2.5 Pro | xhigh | 73.7% |
| 30 | GLM 5.1 | max | 73.5% |