Claude Sonnet 4.5

Modellfamilie von Anthropic · verwendet von 4 Anwendungen

KI-Modell

Claude Sonnet 4.5 is a hybrid reasoning large language model developed by Anthropic, released in September 2025. It excels in coding, agentic tasks, and computer use, with substantial improvements in reasoning, mathematics, and cybersecurity capabilities compared to prior models. The model supports extended thinking mode for complex tasks and is optimized for cost-efficiency while delivering near-frontier performance. Claude Sonnet 4.5 was trained on a proprietary mix of publicly available internet data (up to July 2025), private datasets, opted-in user data, and synthetic data. It underwent RLHF and AI feedback fine-tuning to align with Claude’s Constitution (helpful, honest, harmless). The model is text-only and multilingual. Safety evaluations show strong alignment, with a 99.29% harmless response rate on violative requests and low over-refusal rates (0.02%) on benign prompts. It was deployed under AI Safety Level 3 (ASL-3) protections.