evals.report
BenchmarksLabsCompareRun guides
BenchmarksChat preference

EQ-Bench Creative Writing v3

An LLM-judged creative writing benchmark that scores models across 32 prompts (3 iterations each) using a hybrid of rubric scoring and pairwise Elo comparisons computed with a margin-weighted Glicko-2 rating system.

Chat preferenceEloHigher is better
ScoresAboutRun this benchmark
ModelLabScore↓Source modelStatusDate
Claude Opus 4.7Anthropic2206—VerifiedApr 16, 2026Details
GPT-5.5OpenAI2035—VerifiedApr 23, 2026Details
Claude Opus 4.8Anthropic2031—VerifiedMay 28, 2026Details
GPT-5.4OpenAI2003—VerifiedMar 5, 2026Details
Claude Sonnet 4.6Anthropic1968—VerifiedFeb 17, 2026Details
Claude Opus 4.6Anthropic1933—VerifiedFeb 5, 2026Details
GPT-5.2OpenAI1783—VerifiedDec 11, 2025Details
Kimi K2.6Moonshot AI1782—VerifiedApr 20, 2026Details
Claude Sonnet 4.5Anthropic1767—VerifiedSep 29, 2025Details
Claude Opus 4.5Anthropic1762—VerifiedNov 24, 2025Details
o3OpenAI1744—VerifiedApr 16, 2025Details
Kimi K2 InstructMoonshot AI1738—VerifiedJul 11, 2025Details
Kimi K2 ThinkingMoonshot AI1695—VerifiedNov 6, 2025Details
Grok 4.20 beta reasoningxAI1675—VerifiedMar 9, 2026Details
GLM-5Z.ai1658—VerifiedFeb 11, 2026Details
GLM-5.1Z.ai1645—VerifiedApr 7, 2026Details
GPT-5OpenAI1640—VerifiedAug 7, 2025Details
Claude Opus 4Anthropic1639—VerifiedMay 22, 2025Details
Kimi K2.5Moonshot AI1593—VerifiedJan 27, 2026Details
DeepSeek V4 ProDeepSeek1570—VerifiedApr 24, 2026Details
DeepSeek V4 FlashDeepSeek1556—VerifiedApr 24, 2026Details
DeepSeek V3.2DeepSeek1515—VerifiedDec 1, 2025Details
Claude Sonnet 4Anthropic1514—VerifiedMay 22, 2025Details
Gemini 3 ProGoogle DeepMind1504—VerifiedNov 18, 2025Details
DeepSeek R1DeepSeek1500—VerifiedJan 20, 2025Details
GPT-4oOpenAI1484—VerifiedMay 13, 2024Details
Gemini 3.1 Pro PreviewGoogle DeepMind1479—VerifiedFeb 19, 2026Details
DeepSeek V3 0324DeepSeek1474—VerifiedMar 24, 2025Details
Qwen3.5-397B-A17BAlibaba / Qwen1469—VerifiedFeb 16, 2026Details
Claude 3.5 SonnetAnthropic1448—VerifiedJun 20, 2024Details
DeepSeek V3.1DeepSeek1420—VerifiedAug 21, 2025Details
GPT-4.1OpenAI1419—VerifiedApr 14, 2025Details
Gemini 2.5 ProGoogle DeepMind1417—VerifiedMar 25, 2025Details
GLM-4.7Z.ai1403—VerifiedDec 22, 2025Details
Mistral LargeMistral AI1402—VerifiedFeb 26, 2024Details
Claude 3.7 SonnetAnthropic1395—VerifiedFeb 24, 2025Details
GLM-4.6Z.ai1393—VerifiedSep 30, 2025Details
MiniMax M2.5MiniMax1331—VerifiedFeb 12, 2026Details
Grok 4.1 fast reasoningxAI1317—VerifiedNov 19, 2025Details
GPT-5 miniOpenAI1298—VerifiedAug 7, 2025Details
Reka Flash 3Reka AI1250—VerifiedMar 10, 2025Details
Gemini 2.5 FlashGoogle DeepMind1243—VerifiedApr 17, 2025Details
Gemini 2.0 FlashGoogle DeepMind1239—VerifiedDec 11, 2024Details
GPT-OSS-120BOpenAI1041—VerifiedAug 5, 2025Details
Llama 3.1 405BMeta953—VerifiedJul 23, 2024Details
Llama 4 MaverickMeta927—VerifiedApr 5, 2025Details
Llama 4 ScoutMeta883—VerifiedApr 5, 2025Details

Each row reports the model’s Elo on EQ-Bench Creative Writing v3. Click a row for the full run context.