Claude Sonnet 4.5
The best coding model in the world
About
The page announces the release of Claude Sonnet 4.5, detailing its performance across various benchmarks such as SWE-bench, Terminal-Bench, τ2-bench, AIME, OSWorld, MMMLU, and Finance Agent. It provides specifics on the methodologies used for evaluation, including context configurations, sampling techniques, and the use of extended thinking or tool use. The content also includes links to related news, Anthropic's product offerings (Claude, Claude Code, etc.), different Claude models (Opus, Sonnet, Haiku), solutions for various industries, developer platform resources, learning materials, company information, and legal/support documentation. It serves as a comprehensive update on the new model and a hub for Anthropic's ecosystem.