Anthropic Introduces Claude Opus 4.5 with Strong Gains in Coding and Agents

Claude-Opus-4.5-2

Full size is 1920 × 1280 pixels

Bar chart titled "Software engineering SWE-bench Verified (n=500)" compares accuracy (%) of models: Claude Opus 4.5 (80.9%), Sonnet 4.5 (77.2%), Opus 4.0 (74.5%), Gemini 1.5 Pro, GPT-5.1 Codex-Max, and GPT-5.1.

Bar chart titled “Software engineering SWE-bench Verified (n=500)” compares accuracy (%) of models: Claude Opus 4.5 (80.9%), Sonnet 4.5 (77.2%), Opus 4.0 (74.5%), Gemini 1.5 Pro, GPT-5.1 Codex-Max, and GPT-5.1.