A user on Mastodon expressed significant disappointment with the performance degradation of Anthropic's Opus and Sonnet models over the past 18 months. The user claims that while these models were impressive and useful for reducing technical debt initially, their performance has declined to the point where Opus 5.5 is now considered "borderline not usable" for critical tasks. This perceived decay in model performance is attributed to late 2026, with the user questioning if others have noticed similar issues. AI
IMPACT User feedback suggests potential issues with the reliability and performance consistency of advanced AI models over time.
RANK_REASON User opinion piece about perceived model performance degradation.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →