Anthropic Unveils Claude Sonnet 4.5 – Best AI for Programming and Complex Computing
Credit: Anthropic
Anthropic has introduced a new, cutting-edge model called Claude Sonnet 4.5, which it claims delivers superior performance in programming benchmarks. The company claims that Claude Sonnet 4.5 is capable of creating not only prototypes but also production-ready applications.
Claude Sonnet 4.5 took first place in SWE-bench Verified, an industry benchmark that measures the real-world capabilities of AI models in writing and analyzing code. According to Anthropic, Sonnet 4.5 can maintain concentration for over 30 hours when working on complex, multi-step tasks, outperforming previous versions of Claude and its closest competitors.
In OSWorld’s AI benchmark, which tests real-world computing tasks, Sonnet 4.5 scored 61.4% versus Sonnet 4’s 42.2%, illustrating a dramatic performance improvement in recent months.
The new model has proven itself in more than just programming. According to internal and independent tests, Sonnet 4.5 demonstrates significant improvements in inference and mathematics, as well as in specialized areas such as finance, medicine, law, and STEM. Developers note improved code generation and analysis, file management, and complex real-time calculations.
Anthropic also claims that Claude Sonnet 4.5 is the most advanced AI model to date, with lower rates of sycophancy and deception than previous models. The company also claims that Claude Sonnet has become less susceptible to hint-based attacks.
