NextFin news, On Monday, September 29, 2025, Anthropic, an artificial intelligence company, announced the release of Claude Sonnet 4.5, a new AI model designed to autonomously build software and accomplish complex business tasks. The model is available globally through Anthropic's Claude API and consumer app Claude.ai.
Claude Sonnet 4.5 outperforms previous versions and competitors in software development, computer operation, and task automation. According to Anthropic, the model achieved a 77.2% success rate on the SWE-bench Verified benchmark, which tests real-world software engineering tasks, surpassing the 74.5% of its predecessor, Claude Opus 4.1. Internal tests also demonstrated that Sonnet 4.5 can maintain focus on complex coding tasks for over 30 hours without interruption or performance degradation.
The model's enhanced capabilities extend beyond coding. It shows improved performance in cybersecurity vulnerability detection, financial modeling, research, and forecasting. Anthropic reported that Sonnet 4.5 set a new record on FinanceAgent, a benchmark assessing entry-level financial analyst tasks.
Alongside the model release, Anthropic introduced the Claude Agent SDK, a software development kit that enables developers to build AI agents using Sonnet 4.5's infrastructure. These agents can read and write files, manage long-running tasks, run code, search the web, and coordinate multiple sub-agents simultaneously. This SDK is designed to facilitate the creation of sophisticated AI-driven business automation tools.
Additional product upgrades include new features for Claude Code, Anthropic's coding assistant tool, such as checkpoints to save and revert progress, a redesigned terminal interface, and a native Visual Studio Code extension for smoother integration. A Chrome extension for Max tier subscribers was also launched to enhance accessibility and functionality.
Anthropic CEO emphasized the company's focus on enterprise productivity and competing with AI leaders like OpenAI and Microsoft. The company reported over 300,000 business accounts using Claude Code since its full launch in May 2025, generating significant revenue with a run-rate exceeding $500 million.
Anthropic is also offering a limited-time experience called "Imagine with Claude" for Max subscribers, showcasing real-time software generation by Sonnet 4.5 through a custom user interface. Pricing for Claude Sonnet 4.5 remains at $3 per million input tokens and $15 per million output tokens.
Anthropic recommends upgrading to Claude Sonnet 4.5 for all use cases, highlighting its state-of-the-art performance and extended operational endurance as key advantages in the evolving AI productivity market.
Explore more exclusive insights at nextfin.ai.