CLAUDE OPUS 5.5 DROPS EM DASHES, GETS WORDIER

Analysis of Claude Opus 5.5 reveals a notable departure from writing patterns seen in its predecessor, Opus 5. The model dramatically reduces em dash usage—a hallmark of AI-generated text—while simultaneously increasing response length. The update also shows shorter sentences and simpler vocabulary compared to Opus 5. These changes align with efforts to make AI outputs read more naturally and less like typical machine-generated content. The apparent trade-off between reduced stylistic markers and expanded answers suggests Anthropic may be adjusting how the model balances brevity with comprehensiveness. Whether this reflects intentional design changes or emerges from model training remains unclear. The shift could affect how users perceive Claude's communication style and may influence how AI-generated content is evaluated across different platforms and use cases.