EQ-Bench Longform Writing — leaderboard
Tests sustained creative writing across 8 sequential 1000-word chapters with brainstorming, revision cycles, and 14 quality dimensions including character development and plot coherence.
Metric: Writing Score (0-100). Source: eqbench.com. Status: saturation imminent. 123 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 83 |
| 2 | Claude Opus 4.7 | 81.8 |
| 3 | GPT-5.6 Sol | 81.7 |
| 4 | Claude Opus 4.8 | 80.8 |
| 5 | Claude Sonnet 4.6 | 79.9 |
| 6 | Kimi K3 | 79.6 |
| 7 | Kimi K2.6 | 78.5 |
| 8 | GPT-5.4 | 78.3 |
| 9 | Claude Sonnet 5 | 78.3 |
| 10 | GPT-5.5 | 78.2 |
| 11 | GPT-5.6 Terra | 78 |
| 12 | GLM-5.2 | 77.9 |
| 13 | Claude Opus 4.6 | 77.7 |
| 14 | DeepSeek V4 Pro | 75.6 |
| 15 | GPT-5.6 Luna | 75.5 |
Interactive version: theaggregate.ai/benchmark?slug=eq-bench-longform-writing · How the rankings work · Data refreshed daily, snapshot 2026-07-22.