Atlas

Benchmarks

← All benchmarks

Creative Writing v3 Rubric Score

Chat & Writing · 2025-03-28

This Creative Writing v3 metric grades responses against the benchmark's 20-point writing rubric. Higher point totals indicate stronger creative-writing performance.

Top models (higher is better)

ModelScore
Claude Opus 517.1
GPT-5.517.0
GPT-5.416.9
Kimi K316.9
Claude Fable 516.8
GPT-5 (2025-08-07)16.8
GPT-5.6 Sol16.8
Horizon Alpha16.7
Kimi K2.616.7
Opus 4.816.7
GPT-5.216.7
GPT-5.6 Luna16.6
Opus 4.716.6
GPT-5.6 Terra16.6
Muse Spark 1.116.5
Opus 4.616.5
GPT-5.4 Mini16.5
Sonnet 4.616.5
Sonnet 516.5
Kimi K2 Thinking16.5
DeepSeek-V4-Pro16.4
GLM-5.216.4
Kimi K2 Instruct16.4
Inkling16.4
Opus 4.516.4
Gemini 3 Pro Preview16.3
DeepSeek-V4-Flash16.3
DeepSeek-V3.216.3
o316.3
GLM-5.116.3
Grok 4.516.3
GPT-5.3 Instant16.2
Gemini 2.5 Pro Preview 06-0516.2
Hermes 4 405B16.1
Sonnet 4.516.1
DeepSeek-V3.116.1
GLM 4.616.1
GLM-516.1
MiMo-V2.5-Pro16.1
Opus 416.1
Loading Atlas data…