Anthropic has unveiled its latest frontier artificial intelligence model, Claude Sonnet 4.5, which it described as a major leap in coding performance and reliability. The company said the new model can now build “production-ready” applications, going beyond mere prototypes associated with previous generations of AI. It is accessible via the Claude API and chatbot, and will retain the same developer pricing structure as Claude Sonnet 4 — $3 per million input tokens and $15 per million output tokens.
In recent years, Anthropic has gained a strong following among developers and enterprises due to the robustness of its AI models in software engineering tasks. Big tech firms such as Apple and Meta are said to be using Claude internally, while platforms like Cursor, Windsurf, and Replit have integrated Anthropic’s models through API access. However, the company faces stiff rivalry, particularly from OpenAI’s GPT-5, which has begun outperforming Claude on some key coding benchmarks, intensifying competition in the AI sector.
Anthropic maintains that Claude Sonnet 4.5 delivers industry-leading performance, citing tests on SWE-Bench Verified and reports of its ability to autonomously code for up to 30 hours, even handling tasks like database setup, domain purchases, and security audits. Industry leaders have endorsed the model, with Cursor CEO Michael Truell calling it state-of-the-art for long-term coding projects, and Windsurf CEO Jeff Wang hailing it as a new generation of coding models. The company further claimed it has reduced flaws in alignment, deception, and prompt injection vulnerabilities, making the model its most secure so far.
Alongside Claude Sonnet 4.5, Anthropic also rolled out the Claude Agent SDK, enabling developers to create their own AI-powered agents using the same infrastructure that powers Claude Code. Additionally, it introduced “Imagine with Claude,” a temporary research preview for Max subscribers that demonstrates the AI’s ability to generate software in real time. The launch comes just weeks after Anthropic’s previous release, Claude Opus 4.1, underscoring the breakneck pace of AI development where companies compete to maintain even a brief lead in innovation.