Anthropic launches Claude Sonnet 5.5 with near-Opus performance at half the price

Anthropic released Claude Sonnet 5.5 on Monday, the second model in the 5.5 family after Opus 5.5. The new Sonnet model offers a 30% speed increase over its predecessor, Sonnet 5, while cutting costs by up to 30%. Anthropic positions it as a faster, more affordable alternative to Claude Opus 5.5, excelling at well-defined daily tasks like debugging and crafting polished documents, slides, and spreadsheets.

Early users also noted it improves user interfaces and can generate decks using slide templates with minimal editing. Like Opus 5.5, Sonnet 5.5 boasts clearer, more natural-sounding text. In benchmark tests, it outperforms Sonnet 5 by significant margins, particularly in coding tasks, matching Opus 5.5's performance in most areas.

It scores 52.1% on Cognition's FrontierCode, close to Opus 5.5's 54.4%, and 1,844 on Artificial Analysis' GDPval-AA when compared to GPT-6 Sol's 1,487. While it competes with Opus 5.5 in efficiency and reasoning, Anthropic acknowledges Opus 5.5 still leads in complex, open-ended tasks that require extensive judgment. Pricing remains the same as Opus 5.5, at $2 per million input tokens and $10 per million output tokens, with cache reads priced at $0.20 per million tokens.

Sonnet 5.5 is now accessible on Claude Platform, AWS, Google Cloud, and Azure, with developers needing to update settings for those using thinking mode.