Anthropic has launched Claude Sonnet 4.5, a new frontier model that delivers state-of-the-art performance on coding benchmarks, enabling the development of “production-ready” applications with improved reliability over previous AI models.
The company stated that Claude Sonnet 4.5 is accessible through the Claude API and the Claude chatbot, with pricing for developers remaining unchanged from Claude Sonnet 4 at $3 per million input tokens and $15 per million output tokens. Anthropic’s AI models are utilized by developers and enterprises, with reports indicating that major companies like Apple and Meta use them internally. Additionally, the company sells API access to AI coding applications such as Cursor, Windsurf, and Replit.
This release comes on the heels of recent reports that OpenAI’s GPT-5 has challenged Anthropic’s models by outperforming them on several coding benchmarks. According to Anthropic, Claude Sonnet 4.5 achieves industry-leading performance on benchmarks like SWE-Bench Verified. David Hershey, an Anthropic AI researcher, noted that benchmark results alone do not fully capture the model’s capabilities. Hershey observed that during early trials with enterprise customers, the model coded autonomously for up to 30 hours, building an application, setting up database services, purchasing domain names, and performing a SOC 2 audit.
Michael Truell, CEO of Cursor, described Claude Sonnet 4.5 as having “state-of-the-art coding performance, specifically on longer horizon tasks.” Jeff Wang, CEO of Windsurf, referred to the model as a “new generation of coding models.” Anthropic also claims that Claude Sonnet 4.5 is its most aligned frontier AI model to date, boasting lower rates of sycophancy and deception compared to its predecessors. The company has improved the model’s resistance to prompt-injection attacks.
