Longform Creative Writing
Chat & Writing · 2025-05-23
Longform Creative Writing evaluates planning, revision, and an eight-turn short story or novella from a minimal prompt. Its current v1.11 results use Claude Sonnet 4.6 to report a 0–100 rubric score.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 86.3 |
| Claude Fable 5 | 83.0 |
| Opus 4.7 | 81.8 |
| GPT-5.6 Sol | 81.7 |
| Opus 4.8 | 80.8 |
| Sonnet 4.6 | 79.9 |
| Kimi K3 | 79.6 |
| Kimi K2.6 | 78.5 |
| Sonnet 5 | 78.3 |
| GPT-5.4 | 78.3 |
| GPT-5.5 | 78.2 |
| GPT-5.6 Terra | 78.0 |
| GLM-5.2 | 77.9 |
| Opus 4.6 | 77.7 |
| DeepSeek-V4-Pro | 75.6 |
| GPT-5.6 Luna | 75.5 |
| Kimi K2.5 | 74.9 |
| MiMo-V2.5-Pro | 73.7 |
| GLM-5.1 | 73.5 |
| Opus 4.5 | 73.1 |
| GPT-5 (2025-08-07) | 73.1 |
| GPT-5.4 Mini | 73.0 |
| Inkling | 72.5 |
| Gemini 3.5 Flash | 71.8 |
| GPT-5.2 | 71.6 |
| Sonnet 4.5 | 71.4 |
| Horizon Alpha | 71.2 |
| GLM-5 | 70.9 |
| MiMo-V2-Pro | 70.9 |
| Muse Spark 1.1 | 70.6 |
| Opus 4.1 | 70.5 |
| Kimi K2 Thinking | 69.0 |
| Gemini 3 Pro Preview | 68.8 |
| Grok 4.20 | 68.5 |
| Qwen3.5 397B A17B | 68.3 |
| Gemini 3.1 Pro Preview | 68.2 |
| Kimi K2 Instruct | 68.1 |
| DeepSeek-V3.2 | 66.8 |
| GPT-5 Chat (2025-08-07) | 66.8 |
| Qwen3-Max (2025-09-23) | 66.2 |