Anthropic’s Next-Gen AI Model, Claude Sonnet 4.5, Offers Industry-Leading Coding Performance and Enhanced Reliability
On Monday, Anthropic unveiled its latest advancement in artificial intelligence (AI) technology with the introduction of Claude Sonnet 4.5. The company claims this model delivers top-tier performance on coding benchmarks, marking a significant leap in reliability compared to previous AI models.
Claude Sonnet 4.5 will be accessible via the Claude API and the Claude chatbot. For developers, the pricing remains consistent with that of Claude Sonnet 4, at $3 per million input tokens (approximately 750,000 words, equivalent to the entire Lord of the Rings series). Output tokens will cost $15 per million.
Over the past year, Anthropic’s AI models have gained popularity among developers and enterprises due to their exceptional performance in software engineering tasks. Major tech companies such as Apple and Meta are reportedly using Claude AI models internally, while Anthropic has generated substantial revenue by providing API access to coding applications like Cursor, Windsurf, and Replit.
Recently, OpenAI’s GPT-5 has posed a challenge to Anthropic’s dominance in this sector, outperforming Claude models on multiple coding benchmarks.
Anthropic asserts that Claude Sonnet 4.5 offers leading performance on various coding benchmarks, including SWE-Bench Verified. However, an Anthropic AI researcher, David Hershey, notes that capturing the model’s capabilities solely through benchmark results can be misleading.
During early trials with select enterprise clients, Hershey observed Claude Sonnet 4.5 autonomously coding for up to 30 hours. During this time, the AI model not only built an application but also set up database services, purchased domain names, and performed a SOC 2 audit to ensure the product’s security.
In a statement shared with the media, Cursor CEO Michael Truell described Claude Sonnet 4.5 as delivering state-of-the-art coding performance, particularly for long-term tasks. Windsurf CEO Jeff Wang referred to it as a “new generation of coding models.”
Anthropic also claims that Claude Sonnet 4.5 is its most aligned frontier AI model yet, with reduced rates of sycophancy and deception compared to previous models. The company has also enhanced Claude’s resistance to prompt injection attacks.
In addition to the launch of Claude Sonnet 4.5, Anthropic is introducing the Claude Agent SDK. This infrastructure, which powers Claude Code, can aid developers in building their own agents.
Anthropic is also offering a temporary research preview called “Imagine with Claude” for Max subscribers. This AI model generates software on demand and responds to user requests in real-time without predetermined functionality or prewritten code.
The fierce competition in the AI sector has led companies to release flagship models at regular intervals. Claude Sonnet 4.5 is being launched less than two months after Anthropic’s previous AI model, Claude Opus 4.1. This rapid production cycle makes it challenging for any company to maintain a lasting competitive advantage.
