Anthropic released Claude Sonnet 4.5, his latest model on Monday. The company has positioned it as the best AI coding system in the world and a leap forward in applied artificial intelligence.
The upgrade arrives only four months after its predecessor, Sonnet 4, highlighting the rapid rate of the startup product in the generative arms race on AI.
Anthropic Said Sonnet 4.5 provides advanced results on Swe-Bench verified, a Standard to assess software engineering performance.
The startup has also presented the capacity of the new model to generate practical commercial results thanks to autonomous computer use, cybersecurity capacities and the creation of applications practiced in production and contextual AI agents.
Anthropic’s revenues have increased this year, mainly motivated by the coding features of its models and a specific product called Claude Code. The startup has moved away from competitors in AI coding, and Sonnet 4.5 is designed to maintain this prospective.
Until now, automated and assisted software coding has been one of the most convincing use cases for a generative AI. This is partly because there are large potential productivity gains and cost savings.
Anthropic noted on Monday that Claude Code generates more than $ 500 million in execution rate income, with use of more than 10 times in three months.
The startup said that Sonnet 4.5 improves the reliability of the code, the refactoring judgment and the production of production. This new model competes with other offers such as Google Gemini, Openai GPT-5 and XAI Grok 4.
Commercial implications are wide. In cybersecurity, Anthropic said that Sonnet 4.5 helps detect and correct vulnerabilities faster. In financial services, it exceeds Anthropic’s 4.1 opus, the most advanced reasoning model in society, in tasks such as research, modeling and forecasts.
Perhaps more particularly, the new model can work independently for 30 hours, more than quadruple the endurance of opus 4, said Anthropic.
The startup also extends its developer ecosystem with new tools that provide Claude Code construction blocks to a wider audience. Developers have access to virtual machines, memory and context management.
New features focused on developers include a native Vs code extension, improved terminal workflow and control points that allow engineers to restore the code instantly if their projects fueled by Ai vaient at the track.
On the Claude Developer platform, Anthropic has launched a Claude agent software development kit offering developers with fine grain tools to build personalized AI agents and aware of contexts.
Register for the Bi’s Tech Memo newsletter here. Contact me by e-mail to [email protected].
