Claude Sonnet 4.5 (latest)

Claude Sonnet 4.5 is positioned within the Claude family as a model with a strong emphasis on coding and software engineering tasks. Third-party evaluation places it at 77.2% on SWE-bench Verified, a widely used benchmark for real-world coding ability, giving it a slight edge over competing frontier models on agentic programming workloads. This coding orientation is the clearest distinguishing trait supported by available evidence.

Beyond the benchmark figure, the available sources offer little additional detail about Claude Sonnet 4.5's architecture, training approach, parameter count, context window behavior, or modality design. The model's intended role appears to center on developer-facing coding assistance, particularly for agentic programming workflows, though broader qualitative strengths in reasoning, tool use, or multimodal handling are not substantiated by the supplied excerpts.