Claude Opus 5 Responses Are 3x Longer, Simpler Vocabulary, and Loves Em Dashes
arena · x · 2026-08-13
An analysis of tens of thousands of real-world outputs from Text Arena reveals significant shifts in the writing style of Claude Opus 5:
- Much longer: Averages 510 words per response, over 3x longer than Opus 4.5's 158 words.
- Structurally elaborate: 58% rise in average sentence length and 46% increase in clause frequency.
- Lexically simpler: Long content words dropped by 6.8%, and abstract nouns dropped by 53%.
- Noticeable habits: 2.3x more em dashes, doubled use of phrases like "load-bearing," and 50% more honesty wording like "honestly."
Meanwhile, Fable 5 is 38% more concise than Opus 5 (averaging 316 words) and nearly twice as likely to include praise or validation.
More from Models
- Polymarket Bettors Give Grok 5 Under 70% Chance of Releasing by End of 2026 — Polymarket · 2026-08-13
- Qwen3.8-2.4T Suspected to Be Text-Only Version; Multimodality as a Commercial Edge? — xhluca · 2026-08-13
- MiniMax H3 Settings Comparison: Tested on RTX 3090 Ti with 64GB RAM — Then-Comfortable8258 · 2026-08-13
- Dev Reflects on AI Coding: Models Make Basic Reasoning Errors, Hand-Coding Wins — jsuarez · 2026-08-13
- Quest for Zero Refusal: Do Fully Uncensored AI Models Actually Exist? — luget1 · 2026-08-13
- Grok Hits SOTA on Databricks' OfficeQA Pro V2 Evaluation — GavinSBaker · 2026-08-13