The Chonkerton

Claude Opus 5 Is Highly Capable, But Is No Mythos

ai

Claude Opus five is Anthropic's latest model: half the cost of Fable five, delivering most of its performance on everyday tasks. LessWrong reports that Opus five excels at coding and spreadsheets, with benchmark wins on IMO problems and code generation; Fable remains the only frontier-class option. Where Opus five falters: truly autonomous, large-scale reasoning. The sticking point for users: personality quirks—repetitive phrasing, argumentative tone, occasional negativity—that make it polarizing compared to earlier Claude versions. The upside: exceptional resistance to prompt injection attacks.

Source: https://www.lesswrong.com/posts/Pj4Eewb4KXvXFCcGv/claude-...

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton