--- format: "aidr-story-markdown/v1" id: "25c53dbddd71c5056464e63228c7d4385fe3a14b1fd491ec0a3fa6ed925d4dd8" canonical_url: "https://aidr.today/25c53dbd?lang=en" title: "Anthropic engineer explains why Claude's writing got worse although the model got smarter" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-09-23T15:14:34.000Z" category: "Models" topics: ["anthropic","claude","fine-tuning","reasoning","safety","contrastive-lm","clm-8b","qwen"] source_urls: ["https://the-decoder.com/anthropic-engineer-explains-why-claudes-writing-got-worse-although-the-model-got-smarter/","https://www.marktechpost.com/2026/09/23/contrastive-lm-releases-clm-8b-an-open-system-one-model-that-scores-agent-actions-up-to-9x-faster-than-jev/","https://www.marktechpost.com/2026/09/23/a-coding-guide-to-typesafe-ai-jev/"] summary: "Anthropic employee Jackson Kernion explains why newer Claude models write so oddly. Optimizing for math, code, and technical explanations aimed at other AI models has created a style that sounds like \"overly-dense info dumps\" to humans. Opus 5.5 tries to fix this, but Opus 4.6 remains unmatched as a pure writing model." --- # Anthropic engineer explains why Claude's writing got worse although the model got smarter > [Open the canonical story]() **Published:** 2026-09-23T15:14:34.000Z **Category:** Models **Topics:** anthropic, claude, fine\-tuning, reasoning, safety, contrastive\-lm, clm\-8b, qwen ## Summary Anthropic employee Jackson Kernion explains why newer Claude models write so oddly\. Optimizing for math, code, and technical explanations aimed at other AI models has created a style that sounds like "overly\-dense info dumps" to humans\. Opus 5\.5 tries to fix this, but Opus 4\.6 remains unmatched as a pure writing model\. ## Sources - [Story source]() - [Story source]() - [Story source]()