Anthropic: Claude Opus 4

Anthropic's Claude Opus 4, released May 22, 2025, demonstrates a strong performance profile, particularly in specialized domains. While its speed performance is moderate, ranking in the 30th percentile, its pricing tends to be premium, positioned in the 3rd percentile. A standout feature is its exceptional reliability, achieving a 100% success rate across all benchmarks, indicating minimal technical failures. Opus 4 excels in coding, achieving 94.0% accuracy and being benchmarked as the world's best coding model at release, with leading results on SWE-bench (72.5%) and Terminal-bench (43.2%). It also shows perfect accuracy in Hallucinations, General Knowledge, and Ethics, often being the most accurate model at its price point and speed for these categories. Its Reasoning and Mathematics capabilities are also very strong, scoring 90.0% and 94.0% respectively. Instruction Following is solid at 71.0%. The model's ability to handle extended, agentic workflows for hours without degradation is a significant strength. The primary weakness appears to be its premium pricing across most benchmarks.