Claude Opus 4.5

Claude Opus 4.5: Anthropic's frontier reasoning model, delivering state-of-the-art coding and agentic performance (80.9% on SWE-bench Verified) with new token-efficiency controls.

Claude Opus 4.5 is Anthropic’s most intelligent model, engineered for complex, long-horizon tasks. It sets the performance standard: achieving 80.9% on the SWE-bench Verified benchmark and reducing production code errors by 50-75% (cite: 1.2, 1.4). We’ve integrated the unique 'effort' parameter (low, medium, high) for token-level cost control, optimizing efficiency without sacrificing quality (cite: 1.3). This model dominates autonomous agentic workflows, handles multi-day coding projects in hours, and is available with a 200k token context window for enterprise solutions (cite: 1.3, 1.4). It is the most practical, powerful solution for frontier AI work (cite: 1.3).

https://www.anthropic.com/