Anthropic's Claude Sonnet 5.5 nearly matches Opus 5.5 on benchmarks while costing up to 30 percent less per task
decoder ·
Anthropic has released Claude Sonnet 5.5, the second model in its Claude 5.5 family, which offers output generation speeds more than 30 percent faster than its predecessor while reducing per-task costs by up to 30 percent through improved token efficiency. The new model demonstrates significant performance gains in coding and knowledge-work benchmarks, nearly matching the capabilities of Anthropic's top-tier Opus 5.5 on several tests despite being positioned for everyday tasks like bug fixing and documentation. On benchmarks such as Terminal-Bench 4.0 and GDPval-AA, Sonnet 5.5 shows substantial improvements over Sonnet 5, with scores that closely approach those of Opus 5.5 and OpenAI's GPT-6 Sol. Although it shares the same per-million-token pricing as Sonnet 5, the reduction in tokens required per task results in lower effective costs for users. The model is now available on AWS, Google Cloud, and Azure, featuring adjustable effort settings that allow developers to balance speed, cost, and quality. Notably, Anthropic has introduced new cybersecurity safeguards for Sonnet 5.5 due to its increased capability, including rerouting high-risk requests to Sonnet 5 and adding classifiers against distillation attacks. Early testers have praised the model's natural conversational style and ability to handle complex visual and programming tasks, such as playing Pokémon Red using only screenshots. While independent verification of these claims is still pending, the release intensifies the competitive landscape among major AI providers.
decoder