Anthropic: Claude Sonnet 4

Claude Sonnet 4, released by Anthropic on May 22, 2025, demonstrates a strong overall performance profile, particularly excelling in reliability. It consistently provides usable responses, achieving a perfect 100th percentile in reliability, indicating minimal technical failures. In terms of speed, Sonnet 4 exhibits competitive response times, ranking in the 41st percentile across benchmarks. However, its pricing tends to be at a premium level, positioned in the 12th percentile. The model showcases exceptional accuracy across several key benchmarks. It achieved perfect accuracy in Ethics (Baseline), making it the most accurate model at its price point and speed. Sonnet 4 also performed very strongly in Coding (93.0% accuracy), being the most accurate among models of comparable speed, and in Reasoning (96.0% accuracy) and General Knowledge (99.8% accuracy), both ranking in the top percentiles. While its Instruction Following accuracy (66.0%) is solid, it is not as dominant as its performance in other areas. Email Classification also shows high accuracy at 99.0%. Sonnet 4's key strengths lie in its advanced reasoning, coding capabilities, and ethical alignment, making it suitable for complex software development and agent-driven workflows, as highlighted by its state-of-the-art SWE-bench performance. Its primary limitation is its premium cost.